When Copyright Law and Science Collide: Empowering Digitally Integrated Research Methods on a Global Scale

INTRODUCTION

Much scholarly attention has focused on possible impediments to both science and innovation arising from extensions of patent protection to research tools and to other upstream knowledge assets in ways that threaten to undermine the cooperative norms of basic scientific research.¹ Until recently, however, far less attention has been paid to the growing capacity of global copyright law and related rights to impede access to, and use of, the cumulative scientific literature and data that digitally integrated scientific research methods massively ingest.² In this Article, we contend that this latter phenomenon poses a more immediate and pervasive threat to basic scientific research methods today than the still controversial claims about thickets of rights and anticommons effects attributed to excesses of the patent system in recent years.³

A. Potentially Boundless Scientific Opportunities in the Digital Environment

Information technology is transforming fields as diverse as molecular biology, especially genomics and proteomics,⁴ and conservation ecology,⁵ while spawning new fields, such as metagenomics⁶ and metabolomics.⁷ The combination of massive storage capacity, powerful data manipulation techniques, and graphical capabilities has revolutionized both how basic research is conducted and how the resulting knowledge is preserved and disseminated in nearly all fields of science.⁸ These methodologies have also helped to generate networked communities of users and collaborators, often working in dynamic knowledge hubs,⁹ whose interactive communications steer computational applications in potentially more fruitful directions¹⁰ and fill open repositories with new data and information.¹¹

In this promising new research environment, scientists increasingly rely on automated knowledge discovery tools to mine and recombine vast amounts of data and literature that are flowing at rates that exceed the capacity of a single investigator to comprehend and manage.¹² Exploitation of these new opportunities, in turn, requires integration of information and data scattered over a broad range of articles and databases that may or may not be available online for extensive computational research.¹³ For example, the use of networked computational techniques for linking global collections of articles and data to generate relevant research results, makes it possible to build field-specific knowledge repositories that capture reams of relevant scientific data and technical information and to apply general data-mining tools in the chosen environment.¹⁴ Users receive value when such tools can also be readily applied to the scientific literature.

The digitization of research inputs and outputs has thus engendered opportunities for the enhanced speed of dissemination of publicly funded scientific data, for the development of high performing search engines that diminish the search time for publications, and for automated cross-linking and text-mining based on standardized metadata. The goal of this digital infrastructure should be to maximize these opportunities for public research institutes and universities in both developed and developing countries, while maintaining the classical functions of certification and diffusion of research results of the predigital print markets.

B. Copyright and Related Laws as Digital Gridlock

To make full use of search engines, data-mining techniques, and other automated knowledge discovery tools, scientists need unrestricted access to a broad range of journals and databases, and unrestricted rights to extract, use, and reuse the published research results they contain for purposes of future research.¹⁵ The convergence of computerized technologies and telecommunications networks has now made this goal the-oretically feasible, and the sharing norms of science pull in the same direction.¹⁶ Researchers anywhere should, in principle, be able to locate, analyze, and disaggregate collections of scientific information and data once they have been digitally transmitted and made available to the public, subject only to the prevailing community norms of attributions.¹⁷

In reality, intellectual property laws, as currently configured, stand in the way of attaining these goals. Since the 1990s, in particular, there has been an unprecedented extension of copyright law and related rights protecting both literature and collections of data into the realm of basic science, with no adequate exceptions for research as such.¹⁸ These developments tend to subject the growing profusion of scientific data and information to the same unbridled proprietary impulses that have lately dominated the regulation of creative endeavors in the traditional arts.¹⁹

For example, global copyright laws automatically confer exclusive proprietary rights on authors of scientific literature,²⁰ who routinely transfer those rights to commercial publishers.²¹ Database protection laws, now enacted in more than fifty-five countries, simultaneously endow compilers and publishers (as assignees) with exclusive rights to the very data that copyright laws traditionally left unprotected.²² Publishers, in turn, surround both scientific data and literature with a variety of technological protection measures (TPMs)—so-called electronic fences and digital locks—that cannot be penetrated or pried open even for purposes of scientific research without violating global norms rooted in an array of multilateral, regional, and bilateral treaties, as well as in a host of national legislative and regulatory instruments.²³

The end result is a growing conflict between private rights and public goods at the core of today’s most promising research techniques.²⁴ Enlightened policymakers view these upstream data and information resources as public goods that need to be widely shared in order to produce more downstream commercial applications that advance public welfare.²⁵ In contrast, intellectual property laws now impede access to scientific data and literature, just at the time when developments in scientific research methods require the use of automated knowledge discovery tools that depend on unfettered access and re-use conditions for their successful applications.²⁶

C. Nature and Scope of This Article

This Article examines the complex challenges posed by copyright and related laws for digitally integrated scientific research, which have emerged, piece by piece, from several decades of disjointed legislative initiatives undertaken at the global, regional, and national levels. We explain how this state of affairs came about and why fairly radical legislative reforms would be needed to undo the harm that this tangled regulatory net inflicts on global scientific research. More realistically, we then explore a number of self-help measures that the scientific community could itself adopt to alleviate some of the pressures emanating from a bevy of poorly conceived intellectual property laws. The overall goal is to persuade policymakers to avoid measures that might further fragment and balkanize the research environment and to affirmatively empower the public good functions of these laws once again, with a view to stimulating more and better scientific outputs and more downstream commercial applications.

In Part I, we map the historical context and evolution of the current deep divide between copyright law and science. Given that digital scientific research is necessarily global in its sweep, we focus considerable attention on comparative laws that tend to fragment essential research inputs into diversely accessible territorial compartments and to marginalize the need for unified fields of enquiry. We demonstrate that, under current conditions, scientists using automated knowledge discovery tools will likely become collective infringers of both domestic and international copyright laws and of national database protection laws where applicable.²⁷

In Part II, we assess the limits of incremental legislative or judicial action traditionally associated with copyright reform processes and make the case for a strong and broad exception for scientific research. Such an exception, we argue, must be buttressed by imposing limits on the use of digital locks and related contractual restraints on users, lest publicly funded science become a hostage to the privatization impulses engendered by para-copyright regimes. We further propose complementary reforms to both national database protection laws and international intellectual property standards consistent with the need to empower e-science to flourish in a digitally integrated research space.

Finally, in Part III we argue that the best outcome for the future of scientific research may well be for the scientific community itself to take responsibility for managing the conditions under which its own knowledge assets will be created and deployed. We reconsider the wisdom of continuing to rely on proprietary publishing intermediaries in an environment increasingly characterized by an array of promising open access options. This Article concludes with some final observations on the need to overcome the disconnect between private rights and public scientific research goals as a step towards the elaboration of a more rational, long-term innovation policy.

I. THE GROWING DIVIDE BETWEEN COPYRIGHT LAW AND SCIENTIFIC RESEARCH IN HISTORICAL PERSPECTIVE

Traditionally, copyright law and science operated in two relatively distinct spheres, with patents seen as the primary source of incentives for applications of research results to industry. Although most national copyright laws protected scientific literature,²⁸ this protection narrowly covered the author’s mode of expressing research results, and not facts, data, or ideas as such.²⁹ In a major decision in 1991, the U.S. Supreme Court further truncated compilers’ rights³⁰ by allowing third parties to make use of the disparate facts and data revealed even in an otherwise eligible compilation, notwithstanding the copyright owner’s exclusive right to prepare derivative works.³¹ Since 1994, international copyright law has also cautiously limited protection of so-called factual works—that is, compilations of facts and data—to their original selection and arrangement, but not to the underlying facts and data themselves.³²

Questions about unauthorized reproductions of published research results in scientific journals were typically resolved by limitations and exceptions in the domestic copyright laws,³³ especially a fair use exception in the United States and a private use exception in European copyright laws,³⁴ although these issues became much more complex with the advent of photocopying machines.³⁵ Moreover, U.S. copyright law has long dedicated government-generated data and literature to the public domain,³⁶ a practice that, until recently, had been rejected by many other members of the Organization for Economic Cooperation and Development (OECD).³⁷ What emerged, at least in the United States, was a relatively benign regulatory tradition that was further complemented by the sharing ethos of science, which favors open access to published research results for purposes of verification and the progressive generation of further research.³⁸

This traditional approach, however, has been subverted by much discussed high-protectionist trends evolving in multiple directions.³⁹ For example, in an effort to restrain perceived acts of misappropriation, some federal appellate courts in the United States devised subtle doctrinal arguments to justify greater protection of disparate facts and data than the Supreme Court’s “thin copyright” approach to compilations would otherwise seem to have warranted.⁴⁰ Outside the United States, efforts to strengthen the protection of factual compilations led some fifty-five countries, mostly, but not exclusively, affiliated with the European Union, to enact sui generis database protection laws that deviate from copyright tradition by directly protecting facts and data as such.⁴¹

Meanwhile, both the Agreement on Trade-Related Aspects of Intellectual Property Rights of 1994 (TRIPS Agreement) and the World Intellectual Property Organization’s (WIPO) Copyright Treaty of 1996 (WCT )⁴² have imposed outer boundaries on limitations and exceptions to the exclusive rights recognized in national copyright laws. These laws cast some doubt on the continued ability of prior doctrinal tools to alleviate impediments to the conduct of scientific research.⁴³ Finally, the advent of digital technologies, and the global response to their impact on the transmission of copyrighted works online, has led both the United States and the European Union to adopt regulatory regimes, ostensibly pursuant to the WCT,⁴⁴ that can prevent the use of most existing limitations and exceptions, and even prevent third parties from accessing unprotectible facts and ideas.⁴⁵ As these privatizing trends increasingly encroach on the realm of scientific research,⁴⁶ access to basic knowledge inputs becomes ever more complicated and potentially difficult or costly to obtain.

A. Two Conceptual Approaches in the Application of Copyright Law to Science

The well-known philosophical differences between Continental “authors’ rights” laws, rooted in natural law tradition, including protection of the author’s personality interest, and the copyright laws of common law countries, rooted in utilitarian notions of social welfare,⁴⁷ led logically to contrasting views of limitations and exceptions to the basic bundle of authors’ rights.⁴⁸ Under the Continental tradition, largely embodied in the Berne Convention of 1886, most uses of an author’s creative work presumptively require compensation.⁴⁹ Any exceptions to or limitations on that principle should be narrowly drawn, lest authors be saddled with obligations to finance public goods that were not imposed on other forms of property.⁵⁰ Under the copyright approach, instead, as elaborated most fully in the United States, authors should receive only those entitlements needed to overcome the risk of market failure posed by free-riding copiers, and these entitlements remain subject to carve outs that support the public interest ab initio.⁵¹

These different philosophical foundations produced two different approaches to limitations and exceptions bearing on the exclusive rights that copyright law confers on authors of literary and artistic works. In Europe, the standard approach was to establish a list of enumerated exceptions, with the understanding that activities not covered by any of the listed exceptions were usually proscribed, even if they sometimes appeared to be natural extensionsof an existing exception.⁵² These codified exceptions thus need updating at regular intervals, and they are interpreted narrowly by courts, who tend to view them as undermining the dominant theme of authors’ property rights.⁵³

In contrast, U.S. legislation combines a list of fairly specific express exceptions to the exclusive rights of authors with a broad fair use provision that carves out additional space for noninfringing activity, usually transpiring within specified normative guidelines.⁵⁴ This open-ended carve-out then applies not only to new situations not directly reached by the codified list of exceptions, but it may sometimes retroactively expand even the scope of those exceptions that are codified.

The differences between these two approaches have clearly diminished over time, as policymakers on both sides of the Atlantic rely on both incentives to create and natural-property-rights thinking to justify ever higher levels of copyright protection.⁵⁵ Conversely, scholars in Europe increasingly focus attention on the need for an appropriate balance between protection and free uses.⁵⁶ As will be seen below, a degree of harmonization has also been superimposed upon all the domestic copyright laws of WTO Members by international law. Nonetheless, these historical foundations help to explain the differences that still characterize the distinctive approaches to limitations and exceptions in the European Union and the United States.

1. Harmonizing the Designated Limitations and Exceptions that Defend Scientific Research in the European Union

In Continental Europe, limitations and exceptions to copyright law emerged from state practice and over time were largely incorporated into revisions of the Berne Convention of 1886.⁵⁷ The Berne Convention thus supplied the primary harmonizing platform for limitations and exceptions throughout the twentieth century, even though there still remained some undefined, if contested, space for supplementary state action.⁵⁸ In this context, early exceptions for science were squeezed into Article 10(2) of the Berne Convention, which as late as the Brussels Revision of 1948 allowed States to provide exceptions for “excerpts from literary or artistic works in educational or scientific publications,” but only “in so far as this inclusion is justified by its purpose.”⁵⁹

Even this simple, if rigid, approach (still extant in the copyright law of the United Kingdom),⁶⁰ was not mandatory.⁶¹ States attempting to facilitate science under its aegis need not have followed any particular model or any agreed view of the needs of science as distinct from or constrained by the economic interests of authors and publishers.

Things changed, however, when at the Stockholm Revision Conference in 1967, the Berne Convention formally incorporated an exclusive reproduction right into Article 9(1), along with a three-step test for enabling exceptions to that same reproduction right in Article 9(2).⁶² That test confined national legislation on exceptions to the reproduction right to “certain special cases,” that did “not conflict with a normal exploitation of the work” and that did “not unreasonably prejudice the legitimate interests of the author.”⁶³

Although nothing in this provision dealt expressly with science, the legislative history confirms that it was intended to govern the use of scientific literature for research purposes.⁶⁴ For this very reason, the express reference to science in regard to permissible excerpts under Article 10(2) was deleted at the same time.⁶⁵ A truncated version of Article 10(2), which now only regards excerpts for teaching, was ultimately incorporated into the 1971 Revision of the Berne Convention.⁶⁶ The decision to deal with the principal exception for science under the newly enacted three-step test of Article 9(2), rather than by means of a separate provision, introduced new levels of uncertainty⁶⁷ about the scope of permissible scientific activities without mandating any specific action favoring scientific research as such.

Meanwhile, because the exceptions covered in the Berne Convention still were not exhaustive, a number of states adopted supplementary measures. For example, language ambiguously allowing reproductions and other uses “for the sole purpose of illustration for teaching or scientific research” was adopted in some national copyright laws.⁶⁸ State practice in mostly European countries also recognized a “private use” exception⁶⁹ that enabled scientists to make verbatim copies by hand of literary works for research purposes.⁷⁰ It was this latter provision that effectively promoted scientific research over-and-above other designated exceptions in the Berne Convention or state copyright laws, at least until the advent of photocopying machines, and then digital reproduction technologies, which led to regulation of the use of these devices in the interest of publishers.⁷¹

A major effort to harmonize limitations and exceptions at the regional level then occurred in 2001, with the adoption of the Directive of the European Parliament and the Council of Europe on the Harmonization of Certain Aspects of Copyright and Related Rights in the Information Society (InfoSoc Directive).⁷² Ostensibly devised to implement the WCT of 1996⁷³ and the TRIPS Agreement of 1994,⁷⁴ this Directive sets out a deliberately exhaustive list of permissible exceptions and limitations to the exclusive rights of authors that European Union member states may enact at their discretion.⁷⁵ Besides allowing reproductions for photocopying, subject to payment of fair compensation, and for noncommercial reproductions by public libraries under Article 5(2),⁷⁶ the Directive expressly mentions scientific research in Article 5(3)(a). Echoing some prior state practice, this provision allows “use for the sole purpose of illustration for teaching or scientific research,” so long as the source, including the author’s name, is indicated … and “to the extent justified by the noncommercial purpose to be achieved.”⁷⁷

The meaning of this ambiguous provision is hard to pin down with any degree of certainty. Narrowly, it seems to limit excused uses to cases of “illustration” for both teaching and scientific research, unless the term “scientific research” can legitimately be detached from “the sole purpose of illustration” language. Although some state practice may lean towards such a broader construction favoring reuse of information for further research,⁷⁸ language concerning related rights in the Rome Convention of 1961 avoided any such ambiguity. It excused “use solely for the purposes of teaching or scientific research” without mention of illustrations or a non-commercial purpose as qualifying conditions.⁷⁹ Current applications of the Vienna Convention on the Law of Treaties may make it difficult to ignore this difference in language when courts take the InfoSoc Directive as a standard for construing national laws that implement its provisions, which renders a broad interpretation favoring science less likely to pass muster.⁸⁰

Even if a broader interpretation were to prevail ( by limiting the term “illustration” to exceptions for teaching), it must still overcome the Directive’s noncommercial purpose qualifier.⁸¹ Because universities now routinely engage in commercial exploitation of their scientific research results in both the European Union and the United States, rights holders (typically publishers) can argue that the bulk of such research is commercial in the strict sense of the word. Such an interpretation was recently upheld in a decision concerning university patents by the United States Court of Appeals for the Federal Circuit,⁸² although it is not clear that European courts would take a similarly strict line in regard to either patents or copyrights.

In this unfavorable setting, Article 5(3) of the InfoSoc Directive has done little to strengthen or encourage digital scientific research or the rights of scientific investigators. To the contrary, the Directive may have fatally weakened them by definitively subjecting the old private use exception to a “pay equitable compensation” principle. This conclusion follows because, empirically, there is reason to believe that scientific research in the European Union actually relied on the private use tradition found in most domestic copyright laws, but never directly mentioned in either Berne or TRIPS.⁸³

In any event, the ability of scientific researchers to fall back upon the traditional private use exception had already been compromised by the advent of modern means of technical reproduction, which risked allowing the exception to swallow the exclusive reproduction right and invited countervailing regulatory action in national laws. In this vein, the InfoSoc Directive responds with a double-edged regulatory sword in Article 5(2)( b) by subjecting both photocopying and private copying to an express obligation to pay fair compensation.⁸⁴ Arguably, these provisions cut back upon the preexisting ability of scientists to broadly copy literature for research purposes under the private use exception.⁸⁵ Moreover, apart from certain noncommercial library reproductions, these provisions in effect largely confine scientific research to the vague and somewhat ambiguous language of Article 5(3)(a) as implemented in actual state practice.⁸⁶

Finally, the exhaustive list of permissible exceptions in the EC’s Directive contains no fair use provision that might afford a greater degree of flexibility.⁸⁷ On the contrary, Article 5(5) of the EC’s InfoSoc Directive imposes three additional requirements that negatively circumscribe all the limitations and exceptions it otherwise allows.⁸⁸ This three-step provision—derived from Article 13 of the TRIPS Agreement—embodies a retroactive and excessively narrow reading of the applicable international minimum standards, and thus appears to ignore more flexible language later embodied in the WCT of 1996 and more clearly amplified in the accompanying Agreed Statements.⁸⁹ As we demonstrate in Part I.A.3, the end result is that the InfoSoc Directive, regardless of how it is implemented, could significantly cut back on the already narrow sphere of exceptions favoring scientific research in the past, whether or not this was its intended purpose.⁹⁰

2. The Fair Use Approach in the United States

In contrast, the United States, which did not join the Berne Convention until 1989,⁹¹ adopted a different approach to limitations and exceptions in general and to those bearing on research in particular. The designated limitations and exceptions in the U.S. Copyright Law of 1976 that are most relevant to scientific research include limitations on the reproduction rights for libraries in § 108⁹² and, above all, the fair use exception codified for the first time in § 107.⁹³ By setting out the conditions under which library reproductions and interlibrary loans might be made for purposes of private study, scholarship, or research, § 108 operates in effect as a codified specification of fair use as it applies to libraries in general.

There are no other designated exemptions bearing on quotations, excerpts, or scientific research as such in the 1976 Act, like those under the Berne Convention⁹⁴ and the European Commission’s InfoSoc Directive.⁹⁵ Hence, it is the codified fair use doctrine, as judicially interpreted, that effectively governs the rights of researchers in the United States to avoid or mitigate the exclusive rights of authors and publishers. Any so-called private use exceptions, comparable to those traditionally found in European copyright laws,⁹⁶ must stand or fall as fair uses in U.S. law.

Section 107 of the 1976 Act expressly recognizes a set of preambular uses or “purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research” for which an open-ended fair use exception is deemed particularly suitable.⁹⁷ These uses promote public goods in ways that courts must reconcile with the private rights of authors in an appropriately balanced copyright system. Much depends, however, on how judges determine whether the harm incurred by the copyright owner is justified by the benefit to the public from allowing the use in question.⁹⁸

To answer that question, the statute requires courts to evaluate four separate criteria that may pull in different directions and evince different weights, viz: (1) the purpose and character of the use (such as a noncommercial use or a so-called transformative use); (2) the nature of the copyrighted work (for example, is it of a factual or scientific character to begin with); (3) the amount and substantiality of the portion used (in both quantitative and qualitative terms); and (4) the effect of any given use upon the potential market or value of the copyrighted work.⁹⁹

In the past, and for a fairly long period of time, it was the fourth factor—the so-called market harm test—that predominated in the case law.¹⁰⁰ Following Campbell v. Acuff-Rose,¹⁰¹ however, all four factors must now be weighed by the courts, and the predominant factor has become the first, in practice, as courts focus increasingly on the presence or absence of a so-called transformative use, i.e., a new use not necessarily envisioned by the original author that enriches culture or the pursuit of knowledge.¹⁰² This transformative use factor has prevailed in the digital arena, as seen in cases involving search engines that access and index massive amounts of data and information on the Internet.¹⁰³ Courts in the United States have now routinely held that the use of thumbnail images as markers for search engines, for example, is transformative and that the fair use defense can avail notwithstanding some use for commercial gain.¹⁰⁴

Empirically, it can be demonstrated that judges also invoke other factors, especially a hidden fifth factor—namely, “the extent to which the claimed fair use serves the public interest”—without which few, if any, major federal appellate decisions affirming fair use are likely to be found.¹⁰⁵ In other words, the federal appellate courts look for transformative uses that advance the public interest, especially as identified with those public goods set out in the preamble,¹⁰⁶ without unduly compromising the author’s reasonable expectations of commercial gain. In this context, public scientific research had fared relatively well under the fair use doctrine,¹⁰⁷ at least until the Digital Millennium Copyright Act (DMCA) was adopted in 1998.¹⁰⁸ For example, it was not customary in the United States for scientists to spend scarce research dollars on payments for photocopied articles for private research, as regularly occurs in Europe,¹⁰⁹ notwithstanding the absence of any private use exception in the 1976 Act.

Underlying unresolved conflicts concerning the theoretical groundings of fair use can, nonetheless, affect a court’s willingness to expand or contract the doctrine on a case-by-case basis.¹¹⁰ For example, much depends on whether fair use is seen as a mere technical adjustment allowing infringement of exclusive rights in exceptional cases, or whether it covers areas of use from which authors were granted no such exclusivity to begin with, in which case it can or should be seen as a form of “users’ rights.”¹¹¹

More recently, under the influence of law and economics theory, the extent to which fair use is justified by various forms of market-failure analysis has attracted considerable attention from courts and commentators.¹¹² For example, at least one relatively recent case concerning photocopies for classroom use¹¹³ and another for use at a commercial scientific laboratory¹¹⁴ seemed to presage a growing judicial resistance to fair use, even for research purposes, especially where novel licensing strategies had emerged. In this vein, courts began to reason that yesterday’s market failure—due, say, to the high transaction costs of seeking and negotiating permissions to use in certain cases—might be cured by tomorrow’s establishment of clearinghouses, standard-form contracts, and digitally regulated access controls, which facilitate pay-per-use mechanisms not conceivable in a less technologically sophisticated era.¹¹⁵ Fair use could then depend on the willingness of courts to envision public good overrides that apply irrespective of market failure.

For a time, the influence of market-failure theory, coupled with a surge in the “property rights” approach to intellectual property law generally, elicited growing scholarly criticism of an “incredibly shrinking doctrine of fair use.”¹¹⁶ Fortunately, this period has lately given way to a series of more flexible decisions favoring transformative uses of various kinds and especially those performed by search engines.¹¹⁷ On the surface, these recent decisions would seem more favorable to scientific research, subject to certain inherent constraints limiting the application of the fair use doctrine as a whole. In reality, new constraints arising from technological fencing measures under the DMCA¹¹⁸ can enable content providers to effectively shut down the fair use exception in the online environment, precisely where it is of greatest use to computational science. These measures are discussed below.

B. New Boundaries Imposed by International Law

So far, we have seen that the conceptual approach to limitations and exceptions affecting scientific research in domestic copyright laws differed considerably in the European Union and the United States. Later on, we shall evaluate the strengths and weaknesses of each approach, with specific regard to their implicationsfor science.¹¹⁹ Before doing so, however, we must take account of developmentsat the international level since 1994 that have greatly complicated the practical application of limitations and exceptions under either conceptual approach.

As previously observed, a major turning point had already occurred in 1967, when the Berne Union countries accepted a package deal in which an author’s exclusive right of reproduction was codified in Article 9(1) of the Berne Convention and simultaneously subjected to the “three-step test” in Article 9(2).¹²⁰ Three decades later, history repeated itself when the drafters of the TRIPS Agreement attempted unsuccessfully to negotiate a set of designated limitations and exceptions to the exclusive rights of copyright, patent, and trademark laws that were about to be harmonized in a single convention for the first time. When, at the end of the day, no agreement could be reached on any of the listed proposals under consideration, the drafters made the unprecedented—and some would say mindless—decision to extend that same three-step test, initially adopted as an expedient in 1967, with some minor variations, to all the exclusive rights bestowed on authors, inventors, industrial designers, and trademark owners in the final version of the TRIPS Agreement as adopted in 1994.¹²¹

The three-step test thus became a universal norm of world intellectual property law, binding on some 153 signatories to the Agreement Establishing the World Trade Organization.¹²² Its enforcement also became subject to the WTO tribunals and cross-sectoral remedies governed by the WTO’s Understanding on the Settlement of Disputes (DSU).¹²³

1. NormativeBlindness at the World Trade Organization

One major problem with the three-step formulation as embodied in Article 13 of the TRIPS Agreement¹²⁴ is that it remains devoid of any intrinsic normative guidance. It thus fails to tell courts and administrators what, if any, user pursuits are particularly worthy, from a policy perspective, of affirmative relief from the right holders’ control under the exclusive rights that the TRIPS Agreement obliges WTO members to provide.¹²⁵

So far, the only WTO panel to apply the three-step test of Article 13 dealt with a broad exemption from the public performance rights for radio and television broadcasts of copyrighted musical works in bars and restaurants under § 110(5) of the U.S. Copyright Act of 1976.¹²⁶ After considering both the breadth of the exemption as codified and evidence of its potential substitution effects on live or recorded music covered by Article 11bis of the Berne Convention,¹²⁷ the panel held that § 110(5) was insufficiently “limited” to satisfy the first prong of the three-step test set out in Article 13 of the TRIPS Agreement.¹²⁸

Section 110(5) was accordingly inconsistent with foreign authors’ exclusive public communication rights under Article 11bis of the Berne Convention as incorporated into the TRIPS Agreement via Article 9.1.¹²⁹ While recognizing that “normative considerations” should play a part at step three, and possibly at step two of any analysis under Article 13, the panel’s reified vision of “limited” exceptions under step one¹³⁰ inhibited it from telling us what those considerations might be or how that normative impact should be weighed against rights holders’ interests.¹³¹

To be fair, the WTO panel in the Section 110(5) case found little normative guidance in the legislative history accompanying the original three-step test adopted as Article 9(2) of the Berne Convention at the Stockholm Conference in 1967.¹³² Rather, the experts who drafted that provision produced a single laconic paragraph of explanation, embodied in the Rapporteur’s Statement at Stockholm. According to this source, the three-step test can be read to mean that an objectively small taking of protected matter may be allowed for some purposes, a large taking for any purpose is strongly discouraged, while a medium-sized taking for a good normative purpose might be cured by the payment of equitable compensation.¹³³

The WTO panel may indeed have made this normative blindness even worse when, reasoning from trade law, it asserted that no public purpose was necessary to trigger application of the three-step test to any given dispute under Article 13 of the TRIPS Agreement.¹³⁴ However, the WTO panel’s approach downplayed the fact that the TRIPS Agreement basically deals with private rights.¹³⁵ Even though it constitutes a treaty among sovereign entities, private rights holders are, in effect, a class of third-party beneficiaries, rather like residents of foreign enclaves whose ethnic, linguistic, and educational rights were protected by certain bilateral and multilateral treaties in the past.¹³⁶ Without a public-purpose justification for derogating from the private rights protected under TRIPS, limitations in domestic laws—like those condemned in the Section 110(5) case—could merely allow a state to take money from one private pocket and put it into another.¹³⁷

If the original three-step test embodied in Article 9(2) of the Berne Convention was thus rather normatively blind, that blindness became even more opaque after its incorporation and expansion under TRIPS, Article 13.¹³⁸ For example, there is no express obligation to take third-party interests into account under Article 13, as there is in the corresponding patent law formulation embodied in Article 30 of the same Agreement.¹³⁹ Moreover, the one WTO panel so far convened to consider that formulation in the patent context failed to take into consideration any of the rather evident public health effects of its decision when evaluating step one of the test.¹⁴⁰

Because the formula as thus applied appears normatively blinkered, it tends to give undue positive weight to acquired rights and to codified exceptions recognized in existing legislation, such as the list set out in the EC InfoSoc Directive.¹⁴¹ This approach harbors a flawed methodology because such lists only tell us the results of past legislative compromises. They do not provide a sound normative foundation on which to build, case-by-case in the future, which could free domestic copyright laws from temporal rigidity.

This rigidity is then magnified by the conventional view that, for any given use to qualify as privileged under the three-step test, the decision maker must answer “yes” to all three questions posed by that test.¹⁴² Until recently, this orthodox position went largely unquestioned,¹⁴³ and modified versions of this same approach have been extended to patents, trademarks, and industrial designs. Fortunately, the Max Planck Institute has launched a head-on challenge to this position, as we shall explore in our discussion of possible reforms below.¹⁴⁴

2. Potential Flexibility Under the WIPO Copyright Treaty

With specific regard to copyright law, Article 13 of the TRIPS Agreement expressly extended the three-step test to all the exclusive pecuniary rights covered by the Berne Convention’s 1971 text, as incorporated into TRIPS by virtue of Article 9.1.¹⁴⁵ Article 11 of the TRIPS Agreement also conferred new exclusive rights on authors of computer programs and cinematographic works.¹⁴⁶ On its face, Article 13 could thus be read as a potentially narrowing super-norm applicable to both the extant exceptions in the revised Berne Convention of 1971 and to all preexisting exceptions in the domestic copyright laws of Berne Union members, with perhaps due deference for so-called acquired rights (acquis), that is, certain state practices known to the TRIPS drafters and never expressly challenged or rejected during the Uruguay Round of Multilateral Trade Negotiations.¹⁴⁷

Two years later at the Diplomatic Conference that produced the WIPO Copyright Treaty of 1996, the United States and the European Union submitted a draft text that expressly attempted to codify this narrowing interpretation of the three-step test, which the pending WIPO treaty was supposed to incorporate.¹⁴⁸ However, after extensive debate and the vigorous intervention of the U.S. science agencies and their representatives, a modified approach to the three-step test was enacted in Article 10 of the final text,¹⁴⁹ along with an Agreed Statement unanimously adopted by the parties.¹⁵⁰

Article 10(1) of the WCT states expressly that contracting parties “may” adopt limitations and exceptions to rights granted authors under the new treaty in conformity with the three-step test;¹⁵¹ while Article 10(2) states that these same contracting parties, when applying the Berne Convention, “shall” confine its limitations and exceptions to the three-step test.¹⁵² The Agreed Statement, negotiated with direct inputs from the Presidents of the U.S. National Academies and then Vice-President Gore’s office,¹⁵³ further declared,

It is understood that the provisions of Article 10 permit Contracting Parties to carry forward and appropriately extend into the digital environment limitations and exceptions in their national laws which have been considered acceptable under the Berne Convention. Similarly, these provisions should be understood to permit Contracting Parties to devise new exceptions and limitations that are appropriate in the digital network environment.

It is also understood that Article 10(2) neither reduces nor extends the scope of applicability of the limitations and exceptions permitted by the Berne Convention.¹⁵⁴

These changes were made for the specific purpose of avoiding posterior challenges to the fair use provisions of the U.S. Copyright Act of 1976 (as some scholars had previously feared could happen under the TRIPS Agreement)¹⁵⁵, but whose validity had not otherwise been compromised during the negotiations that led the United States to adhere to the Berne Convention in 1989.¹⁵⁶

The full implications of this complex set of provisions remain to be worked out at both the national and international levels.¹⁵⁷ For example, these provisions could trump arguments that the three-step test can retroactively undo limitations and exceptions existing in national laws prior to the conclusion of the TRIPS Agreement. What cannot be denied is that any future interpretation of the three-step test codified in Article 13 of the TRIPS Agreement must also reflect the posterior, more flexible gloss on that same test as adopted by essentially the same parties two years later, in Article 10 of the WCT and its Agreed Statement.¹⁵⁸

By the same token, these cumulative legislative enactments oblige the members of the WTO to take account of the three-step test whenever they adopt or enforce limitations and exceptions in their domestic copyright laws, under whatever conceptual approach they prefer.¹⁵⁹ Failure to do so could result in adverse legal challenges at the WTO, with the risk of having to pay damages to states whose bargained-for trade expectations suffered harm by dint of any failure of any given exception to satisfy this test.¹⁶⁰

C. The Shrinking Realm of Scientific Users’ Rights Under Either Approach

Seventy years of practical experience with the teachings of the legal realists¹⁶¹ has demonstrated, at the very least, that the quest for legal certainty¹⁶² cannot be separated from a clear analysis of both the underlying purposes of any given set of laws and the empirical experience accumulated over time from judicial efforts to implement those same laws in actual cases.¹⁶³ With specific regard to copyright law, history also teaches that changing factual conditions constantly elicit a need for creative new solutions that are inherently slow to materialize, and destabilizing to boot, while the lobbying of special interests often remains indifferent to the larger public interest that was supposedly to be advanced by any given set of enumerated exceptions.

For example, even the most enlightened policymakers found themselves confronted by a rapidly changing world in which legal categories designed for the print media collapsed or converged in the digital online environment.¹⁶⁴ Just when traditional scientific research methods were overtaken by the rise of computerized information technologies requiring unfettered access to and use of data and information, researchers discovered that the old categories of exceptions inherited from the print media were so freighted with narrow, legalistic interpretations that they could not readily be updated or accommodated to the challenges of modern science, even with the best of will.¹⁶⁵

1. Impeding Scientific Research Even in the Print Media

In this Section, we first evaluate the existing approaches to limitations and exceptions affecting copyrighted scientific research results as they have worked, or failed to work, in the print media for which they were originally designed. We then show how this fragile, predigital foundation was further undermined by efforts to strictly regulate transmissions of copyrighted works over the Internet, and by the database protection laws adopted in the European Union after 1996.¹⁶⁶

a. Strengths and Weaknesses of the Designated Exceptions Approach

The InfoSoc Directive adopted by the European Commission in 2001 was ostensibly an attempt to reconcile the “designated exceptions” approach of prior Continental copyright laws with the provisions of both the TRIPS Agreement of 1994 and the WCT of 1996.¹⁶⁷ It was also supposed to harmonize the limitations and exceptions found in all the copyright laws of European Union member states, as well as those of affiliated and would-be member countries.¹⁶⁸ In reality critics complain that, with respect to harmonization, the Directive failed in its essential purpose,¹⁶⁹ while its response to the WCT digital copyright provisions embodies a high-protectionist regime inconsistent with the balancing norms actually adopted in that treaty.¹⁷⁰ Whether or not these criticisms prove valid, the inadequacies of this Directive with respect to scientific research remain painfully clear.

For example, even though Article 5(3)(a) of the Directive adopts a soi disant exception for science at the regional level that does not appear in the Berne Convention,¹⁷¹ its tortured language may impose qualifiers that cut back upon the provisions of that Convention, which largely leave science at the mercy of the three-step test. Moreover, the exception for science in Article 5(3) is not mandatory, whatever its meaning turns out to be.¹⁷²

If a European Union member country ignores Article 5(3)(a) of the Directive, it remains bound by Articles 10(1) and 10(2) of the Berne Convention, as revised in 1971, which make no mention of science whatsoever.¹⁷³ Even with regard to quotations or to other excerpts for teaching materials, Article 10(1) limits the former to “fair practice … and [the] extent justified by the purpose,”¹⁷⁴ while Article 10(2) permits the latter only “to the extent justified by the purpose … by way of illustration in publications … for teaching, provided such utilization is compatible with fair practice.”¹⁷⁵ These provisions have been narrowly construed in the United Kingdom.¹⁷⁶ while in major markets outside the European Union, such as Brazil, university libraries reportedly ignore the analogous provision in the domestic copyright law altogether and deny students the possibility of photocopying even short excerpts from scientific texts.¹⁷⁷

If, instead, a European Union member state decides to codify the permissible exception for science as set out in Article 5(3)(a) of the InfoSoc Directive, would-be users of scientific works are immediately confronted with the inherent ambiguities of the language in this provision that were identified earlier in this study.¹⁷⁸ For example, do the words limiting use for purposes of illustration pertain to both teaching and research, or are they confined to teaching only? Even if the latter, more liberal interpretation were to prevail, research uses would remain subject to both the non-commercial purpose clause and to the prevailing judicial tendency to view all exceptions and limitations as subordinate to the dominant purpose of copyright law, which in the European Union, is to protect the private interests of authors and publishers.¹⁷⁹

This restrictive tendency is then reinforced by Article 5(5) of the InfoSoc Directive, which makes the three-step test expressly applicable to all the exceptions set out in Article 5.¹⁸⁰ As noted, traditional European jurisprudence requires that all three steps must be applied cumulatively and all must be given an affirmative response if any given act is to avoid a claim of infringement.¹⁸¹ In sum, under the existing regime, nothing compels a European Union member state to favor scientific research, and those policymakers willing to do so encounter a daunting battery of constraints,¹⁸² even without factoring in the effects of special interest lobbying.

At least one expert on the three-step test nonetheless contends that both European courts and legislators could apply it in a more flexible, open-ended manner, somewhat comparable to fair use in the United States, despite the implicitly uncompromising approach of the InfoSoc Directive.¹⁸³ To this same end, the review of the legislative history pertaining to Article 9(2) of the Berne Convention by the Section 110(5) WTO tribunal did uncover a potentially more flexible interpretation than that found in the decisions of many Continental courts.¹⁸⁴ This approach, if subsequently upheld, might have the advantage of supporting greater use of liability rules to resolve hard cases in the print media, which would constitute a welcome addition in many sectors.¹⁸⁵

However, even liability rules, i.e., “take and pay” rules,¹⁸⁶ do surprisingly little to support digitally integrated research methods, which ingest and mine massive amounts of data and information. Attempting to track such uses for purposes of calculating royalty streams conjures up a vision of thickets of rights and boundless transaction costs that could swallow the most robust research budgets, even if some system of automatic micro-payments were to be devised.¹⁸⁷

If anything, the implicit duty to pay equitable compensation that could emerge from a more flexible judicial scrutiny of the three-step test only reinforces the express mandate to pay equitable compensation for so-called private uses mandated by Article 5(2)( b) of the InfoSoc Directive, and vice versa.¹⁸⁸ Taken together, Articles 5(2), 5(3), and 5(5) of the Directive cumulatively discourage, rather than enable, free access and reuse of published scientific information and data by the public research community, even though such research is normally funded by governments or other public entities and made available by researchers to advance knowledge (and their reputational interests), but not for financial gain.¹⁸⁹

All the rigidity and uncertainty inherent in the provisions analyzed above are further compounded by the absence of any true fair use provision in the relevant international treaties,¹⁹⁰ in the InfoSoc Directive, or in the domestic copyright laws of the European Union.¹⁹¹ Besides enabling courts to infuse greater “play in the joints” to facilitate scientific research, as explained above,¹⁹² an explicit fair use provision in international or regional instruments could induce both courts and legislatures to give more weight to Articles 7 and 8 of the TRIPS Agreement, which expressly promote the public-interest goals of copyright law,¹⁹³ and to the express provision favoring scientific research and education in the preamble to the WCT itself.¹⁹⁴

Lacking any such fair use provision, some European courts concerned about excesses of copyright protection have recently felt obliged to invoke human rights, especially constitutionally protected fundamental rights to free speech, as counter-weights.¹⁹⁵ But such well-intentioned judicial ploys introduce significant uncertainty into an already complicated legal scenario,¹⁹⁶ and they could destabilize even well-settled principles of copyright law without directly addressing the needs of scientific researchers as such. One virtue of the fair use provision in U.S. copyright law is precisely its ability to foster a penumbra in which public good uses of protected matter may occur within copyright law itself,¹⁹⁷ while blunting the temptation to reach for more fundamental or constitutional justifications that could produce unintended destabilizing effects.¹⁹⁸

Given these premises, any limitations favoring scientific research that do manage to emerge from the InfoSoc Directive as applied by the domestic laws of European Union member states will likely remain unmanageable and unpredictable for the foreseeable future. Hence, recent proposals to reform the three-step test itself have attracted considerable attention from scholars concerned about the prospects for digitally integrated research methods, as will be discussed below.¹⁹⁹

Short of that, the ineluctable conclusion that emerges from this analysis is that researchers in general, and digitally empowered scientists in particular, will face the dismal prospects of choosing either to desist from pursuing promising research projects, with enormous opportunity costs, or to carry on in the knowledge that they are likely to infringe copyright laws at every turn.²⁰⁰ Needless to say, this choice becomes even starker when the European Union’s database-protection laws are factored into the equation.²⁰¹

b. Limits of the Fair Use Approach

At least in theory, the fair use exception enables courts to address fact patterns that legislators did not, or could not, have foreseen at the time any given copyright law was enacted. This property helps to keep needed responses to public interest priorities up to date without recurring amendments to existing statutes. In this capacity, fair use buttresses both the ideaexpression principle (now codified in the TRIPS Agreement)²⁰² and constitutionally protected free speech in the United States, which in itself tends to favor access to published scientific research as a purveyor of unprotected facts and discoveries.²⁰³ To this end, the preamble to § 107 of the U.S. Copyright Act of 1976 explicitly recognizes scientific research as a privileged subcategory within the general fair use framework.²⁰⁴

i. Inherent Methodological Uncertainties

Nevertheless, whenever any given scientific work becomes the subject of a litigated fair use enquiry, courts must move beyond the preamble and subject it to the four “balancing subtests” codified in § 107, as previously described.²⁰⁵ On the whole, the four-step test of U.S. fair use law accommodated conventional scientific research methods fairly well in the past, especially in view of the proresearch bias specified in the preamble. With few exceptions, uses for scientific research were usually permitted,²⁰⁶ and this tradition played some of the role that a private use exception had performed in Europe.²⁰⁷

To the extent that photocopying became a problem for the case-by-case approach of § 107, industry-wide rules governing library photocopying were negotiated and codified in § 108 of the Copyright Act of 1976.²⁰⁸ Under these provisions, neither students nor researchers are normally charged for library copies made for even large research projects (unlike the situation in the European Union), although charges will apply to copies made for classroom distribution, usually under blanket licenses from a collection society known as the Copyright Clearance Center.²⁰⁹

Occasionally, the federal appellate courts have rendered decisions circumscribing the reach of fair use even with regard to conventional scientific research methods.²¹⁰ But these cases usually turned on the inherently unstable line between commercial and noncommercial research purposes and on the growing ability of technological measures to overcome preexisting market failures.²¹¹ For example, researchers at a petroleum company were obliged to pay for additional copies of back-dated scientific articles available from a collection society, despite their subscription to the journal in question.²¹²

However, subsequent jurisprudence—especially at the federal appellate level—may correct questionable decisions. The very dependence of the fair use doctrine on unique sets of facts makes it relatively easy for later courts to ignore the appearance of stare decisis by relying on subtle distinctions in the underlying fact patterns.²¹³ This adds to the overall flexibility of the U.S. approach.

Critics of the fair use doctrine tend to emphasize the relative uncertainty likely to attend any new fact pattern²¹⁴ and the high transaction costs arising from the need to litigate close cases.²¹⁵ Both objections have some merit but are easily over-stated.²¹⁶ The fair use doctrine cannot intrinsically claim the kind of legal certainty apologists for the positivist approach demand, disregarding their failures to achieve it in practice.²¹⁷ Because fair use decisions are always fact-specific, attorneys can only base their predictions on general categories of fair use as established by the precedents.²¹⁸ Yet, as scholars have shown, when cases are bundled within specified subject-matter categories,patterns of stability and predictability emerge.²¹⁹

Admittedly, any preexisting set of fair use precedents may be enhanced or narrowed by the judicial zeitgeist prevailing at the time. This observation holds particularly true for swings in the judicial pendulum from a more proproperty rights outlook to a more procompetition outlook, and back again, as routinely occurs in U.S. intellectual property jurisprudence.²²⁰ Care must accordingly be taken when predicting a future outcome on a past set of precedents to allow for shifts in the underlying trends, especially with regard to federal appellate and Supreme Court decisions.²²¹ For example, some important older precedents lost their relative weight in the late 1980s when the market failure approach temporarily squeezed other normative considerations out of the fair use equation.²²² But, as we noted earlier, that trend itself has given way to the rebirth of a so-called transformative use doctrine²²³ and its aggressive extension in recent decisions to search engines.²²⁴

At least two commentators who dislike the uncertainty associated with fair use have expressed appreciation for a “public interest” criterion instead.²²⁵ Yet, that criterion harbors a considerable degree of ambiguity all its own (given that copyright law itself expresses one facet of the public interest).²²⁶ Besides, as earlier noted, the federal appellate courts in the United States almost invariably invoke an uncodified public interest criterion when evaluating the express normative factors set out in § 107 of the 1976 Copyright Act.²²⁷ Where science is concerned, moreover, both the public interest criterion and the normative criteria of § 107 should typically favor use and reuse of research results in close cases concerning research methods.

A more troubling source of uncertainty arises from the tendency of U.S.-style fair use decisions to operate under an all-or-nothing premise that the challenged use must either infringe or not infringe, in which case noninfringing uses imply no corresponding compensatory burden.²²⁸ This practice leads some courts to hand down vacillating decisions depending on how they evaluate the appearance of free riding in fact specific situations.²²⁹

In this respect, the three-step test embodied in the TRIPS Agreement has something to teach U.S. courts and policymakers.²³⁰ As acknowledged by a WTO tribunal, the legislative history concerning this test may allow equitable compensation—a “take and pay” rule—to resolve hard cases where more than a little was taken for a particularly valid public purpose.²³¹ United States fair use law might benefit from this additional element of flexibility in close cases, even as European Union law needs the flexibility of fair use, a topic to which we shall return below.²³²

These very sources of uncertainty can, in turn, exert a restraining effect on both publishers and would-be users of copyrighted works, depending on the facts at issue in any given case. While scientific entities must be wary of engaging in litigation that could issue in an adverse precedent for future research prospects, publishers must be equally wary of challenges to the status quo that may further abridge the scope of yesterday’s exclusive rights. This inherent burden of reciprocal uncertainty tends to reduce transaction costs over time by discouraging overly adventurous fair use challenges as potentially too costly for either side.²³³

In this standoff, much fair use occurs by default, although a threat of future litigation acts sometimes as a sword of Damocles. If scientists are frequently wholesale infringers even with respect to conventional research methods, in the sense that they refuse to allow unreasonable or obsolete laws to obstruct customary or necessary research and teaching practices, rights holders have traditionally been reluctant to sue their sources of publication (often their customers as well) in cases where damages appeared modest at best.²³⁴

Nevertheless, the high costs of litigation, combined with recent extraordinary statutory damages awards,²³⁵ may at times deter would-be users from pursuing even meritorious disputes that ought to enable courts to clarify the proper boundaries of a fair use exception.²³⁶ Users may thus find themselves unable to defend their legitimate rights unless some larger nonprofit entities can be found to finance the campaign and absorb its potential costs if the cause is lost. The power of publishers’ cease-and-desist letters in this regard adds to this potential in terrorem effect by warning that certain cases will lead to high legal transaction costs regardless of the merits.²³⁷ At the moment, there is no expeditious, low-cost means of testing a fair use defense without incurring such litigation costs, although several proposals to remedy this defect have been put forward.²³⁸

ii. Outer Limits of the Case-by-Case Approach

None of these considerations addresses the deeper problems that render the fair use doctrine of little use to practitioners of the digitally empowered, computerized research techniques of primary concern in this Article. The systematic need that researchers, as users of automated knowledge discovery tools, have to survey vast or, indeed, unlimited amounts of literature and data in virtually every contemporary, large-scale scientific investigation, particularly in the life sciences, overwhelms the boundaries set by the four-step test of § 107 and increasingly makes a mockery of the very concept of fair use.

Consider, for example, that the implicit purpose of the substantiality test set out in § 107(3) is to ensure that fair use reproductions of a protected text will be quantitatively and qualitatively reasonable in relation to the work as a whole. In no area, not even parody,²³⁹ can this provision be interpreted to permit wholesale reproduction (as technically defined) of every relevant text in every relevant case, which routinely occurs in computational science or in any scientific research project where automated knowledge discovery tools are employed.²⁴⁰ By the same token, the market-harm test of § 107(4)²⁴¹ becomes drained of precedential meaning if the scientific texts thus scrutinized were published to serve both the research needs of the scientific community and the commercial interests of publishers.

Professor Matthew Sag’s brilliant article on copy-reliant technologies²⁴² sheds considerable light on this conflict of interest. His efforts to reconcile the search engine cases with prior decisions concerning transformative uses of copyrighted works under § 107(1) leads him to posit that nonexpressive, nonsubstitutional uses, in conjunction with copy-reliant technologies, should normally qualify as fair uses across the board, especially if the technologies in question were geared to recognize and implement an opt-in clause.²⁴³ When this intriguing proposition is applied to digitally integrated scientific research methods, however, it reveals a number of key differentiating factors.

For example, one must immediately confront the possibility that, from a rights holders’ perspective at least, massive copying of published research articles to generate further research by means of automated knowledge discovery tools colorably represents both a substitutional and an expressive use of those same articles. Even if that were precisely what scientists qua authors most dearly desired in their relentless pursuit of reputational benefits, gratis fair use on this scale is hardly consistent with the aims of commercial science publishers.²⁴⁴

If only scientific researchers were involved as both creators and users of their own published outputs, then Professor Sag’s default formula for fair uses in regard to copy-reliant technologies could significantly improve the research community’s technical legal position, especially if it were underpinned with an opt out, rather than an opt in default condition. Scientists inclined to opt out of such a voluntary pool would immediately incur countervailing peer pressure and perhaps risk jeopardizing future grants to boot. If, instead, scientists constitute the market for published scientific research, and if that published research cannot be freely and digitally perused without impermissible market harm to publishers, then automated research tools risk becoming instruments of massive and systematic infringement, which no transformative use doctrine could excuse if publishers’ customary interests are to be preserved.

That, indeed, poses one of the fundamental questions raised by our present enquiry, namely, should scientific publishers’ customary interests be preserved at the expense of scientists’ need for wholesale access to, and reuse of, the exploding universe of published scientific literature and data? That question, in turn, raises ancillary questions about what added-value the scientific community obtains from its traditional reliance on external, for-profit publishers, and what the opportunity costs would be if the scientific communities were to break that tie to the publishing industry. These and related questions will be more directly addressed in the final Part of this Article.²⁴⁵

For present purposes, what seems undeniable is that the case-by-case approach of the fair use doctrine is currently overwhelmed by the magnitude and scope of copying necessitated by today’s digitally empowered research techniques.²⁴⁶ A proper response to this challenge could require at least industry-wide negotiations and settlements, like those between research libraries and publishers that gave rise to § 108 of the 1976 Act.²⁴⁷ At best, new institutional arrangements would redefine (or eliminate) publishers’ interests in order to satisfy the scientific community’s need for open access to publicly funded research results in any form they were made available to the public.²⁴⁸ Between these two poles, some possible incremental legislative reforms are skeptically examined below.

As matters stand, however, it is the publishers’ lobby that has driven copyright law and policy with regard to science since the 1990s.²⁴⁹ Rather than supporting the needs of digitally integrated scientific research, publishers have managed to surround scientific information and data transmitted online with legally impenetrable electronic fences and with codified database protection laws that threaten the very foundations of contemporary scientific methodology, as explained in the next two Sections.

2. The Coup de Grâce: Digital Locks and Database Protection Laws

A growing number of scientific journals are now published online, without distribution in print copies at all.²⁵⁰ Once digitally transmitted online, researchers anywhere can, in principle, locate, analyze, and disaggregate any collection of scientific information and data that has been made available to the public, subject only to the prevailing default rules of applicable intellectual property laws and to the contractual restrictions that publishers or providers may otherwise impose.

Even when a journal continues to be published in print copies, the articles it contains, along with the supporting data, may often be made available online for the convenience of later scientific researchers.²⁵¹ Although, in principle, these developments greatly facilitate the traditional sharing norms of science, researchers can legally convert analog contents to digital formats only if they remain within the confines of the limitations and exceptions to the reproduction rights discussed above. Under the European Union’s InfoSoc Directive, this process could be prohibitively expensive, as we have seen. Even under the U.S. fair use provision, such quantitatively large amounts of copying for research purposes could result in publishers’ demands for additional compensation, not to mention the costs associated with the copying itself.

Regarding access to and use of information and data made directly available online, whether or not initially published in hard copies, researchers have discovered that virtually none of the proscience measures of copyright law that still survive in print media apply to works transmitted through telecommunication networks.²⁵² The digital revolution that created such promising opportunities for scientific research also generated intense fears that publishers would become vulnerable to massive infringements online and to other threats of market failure.²⁵³ In response, publishers persuaded legislatures to recast and restructure copyright law in the online environment so as to preserve business models built around the print media.²⁵⁴

These new laws make it difficult to trigger preexisting limitations and exceptions in the online environment, including those favorable to science;²⁵⁵ and they enable publishers to embed pay-per-use machinery among other restrictions into electronic fences surrounding online transmissions of scientific articles.²⁵⁶ The most fundamental postulates of so-called users’ rights, such as the idea-expression dichotomy or fair use in the United States, may thus be entirely overridden by a combination of technical protection measures (TPMs), statutory cutbacks, and contractually imposed restrictions rooted in these same provisions.²⁵⁷ In the European Union, moreover, sui generis database protection laws, mandated by the European Commission in 1996, further restricted access to the very facts and data that are the lifeblood of basic scientific research.²⁵⁸

a. Virtual Elimination of Limitations and Exceptions Favoring Science in the Online Environment

The WIPO Copyright Treaty of 1996 (WCT ),²⁵⁹ which established new rules governing digital transmissions of copyrighted works, did not mandate the radical change of the legal infrastructure just described. On the contrary, the WCT reflects a relatively balanced compromise that resulted from the negotiations of stakeholder coalitions with fairly equal bargaining power on both the publishers and users’ sides.²⁶⁰ The preamble itself thus recognizes “the need to maintain a balance between the rights of authors and the larger public interest, particularly education, research and access to information.”²⁶¹

However, the WCT said nothing about how states should implement the anticircumvention norms that defend electronic fences surrounding works transmitted online so as to preserve public interest privileges and immunities. When the treaty was translated into the domestic laws of the United States and European Union, powerful publisher interests persuaded the respective legislatures largely to ignore or override the safeguard provisions otherwise available.²⁶²

In the Digital Millennium Copyright Act of 1998 (DMCA),²⁶³ for example, the U.S. Congress conditioned the ability of third-party users to invoke public interest measures, such as the idea-expression dichotomy or fair use, on their having first gained lawful access to the work being transmitted online.²⁶⁴ Yet, the moment a would-be user seeks to gain lawful access to the copyrighted work transmitted online, he or she will normally encounter one-sided electronic contracts of adhesion that strip away most or all of the public interest user rights nominally available from the domestic copyright law.²⁶⁵ The DMCA thus arguably created a new exclusive “right of access” subject to virtually no preexisting privileges or immunities of interest to scientific users whatsoever.²⁶⁶

A similar state of affairs (with different nuances in different jurisdictions) arises in the European Union. Article 6 of the InfoSoc Directive of 2001 expressly enables domestic legislators to authorize Technical Protection Measures (TPMs) that curtail or override the preexisting limitations and exceptions otherwise available in the hard copy format.²⁶⁷ Article 6(4) of the same Directive then piously admonishes member states “to ensure that right holders make available to the beneficiary of an exception or limitation provided for in national law … the means of benefiting from that exception or limitation.”²⁶⁸ In practice, however, the Directive provides member states with no legal basis for implementing the thrust of Article 6(4), and national legislation concerning TPMs so far tends to largely ignore Article 6(4) altogether, with a few exceptions.²⁶⁹

As a result, technological fencing devices, coupled with electronic contracts, known respectively as TPMs and Digital Rights Management tools (DRMs), enable publishers to automatically protect both data and information delivered through online networks without gaps in enforcement and without any traditional exceptions for science or other public interest purposes.²⁷⁰ When these technological fences and electronic contracts are further supported by anti-circumvention measures that forbid decryption or other means of cutting through such fences,²⁷¹ the publisher’s control becomes virtually absolute. Database protection laws enacted in the European Union can then make this absolute control virtually perpetual to boot.

b. Exclusive Rights in Noncopyrightable Collections of Data

Compilations of facts and data receive relatively thin protection from the copyright laws of both the United States and the European Union.²⁷² Under these laws, only a creative selection and arrangement of facts or data qualifies as eligible subject matter, and the disparate facts remain available for use by third-party compilers,²⁷³ at least in principle, if not always in practice.²⁷⁴ In a remarkable further development, the U.S. government managed to codify both the idea-expression dichotomy and the principle of limited protection for factual compilations, of crucial importance to science, in the TRIPS Agreement²⁷⁵ and the WIPO Copyright Treaty.²⁷⁶ Global copyright law thus, in effect, encourages states to protect so-called factual works against little more than wholesale duplication of an otherwise creatively organized compilation of facts or data, but not the underlying facts or data as such.

In 1996, however, when promulgating its Directive on the Legal Protection of Databases,²⁷⁷ the European Commission took the unprecedented step of enacting a law that established exclusive rights in the very data that copyright laws had left freely available in the public domain.²⁷⁸ Ostensibly motivated by the Commission’s stated goal of increasing the European Union’s share of the global market for directories and compilations in general,²⁷⁹ which so far has proved unattainable,²⁸⁰ this sui generis regime introduced radical new restrictions on access to and use of compilations of data that were previously unknown to any intellectual property paradigm.

For example, no element of originality or creativity is required to qualify for this form of protection.²⁸¹ Instead, the database laws are triggered by a “substantial investment” in obtaining, verifying, or presenting any given collection of facts and data; and unlike copyright or patent laws, the exclusive rights to extract or reuse the data in question protect that investment as such.²⁸² Despite its anomalously low threshold of eligibility, this regime arises automatically, as if it were part of the copyright infrastructure. It thus poses a direct threat to digitally integrated scientific research by endowing compilers of noncopyrightable collections of data with exclusive rights to extract and reuse the disparate data that their sweat-of-the-brow investment made available to the public.²⁸³

These exclusive rights to data are potentially stronger and more rigid than those of copyright law.²⁸⁴ Formally, independent creation remains a perfect defense,²⁸⁵ as it would under copyright law.²⁸⁶ Realistically, however, independent generation of costly accumulations of scientific data is economically unfeasible, even when conceptually possible.²⁸⁷ The Directive does allow a “small” amount of data to be taken without consequence, but courts have ensured that “small” means very small, and the Directive expressly prohibits repeated extractions of even small amounts of data from the same collection.²⁸⁸

Permissible exceptions to the database regime are paradoxically truncated when compared with those of copyright law.²⁸⁹ With specific regard to the use of protected data for scientific research, the Directive allows states to adopt an exception couched in the same ambiguous language as that of the InfoSoc Directive of 2001, namely, “for the sole purpose of illustration for teaching or scientific research.”²⁹⁰ As in the InfoSoc Directive, this exception is not mandatory, and major countries such as France and Italy have ignored it.²⁹¹ Even when countries adopt this exception, it seems to enable only extractions for purposes of illustration, but not for reutilization of scientific data or information in other collections, which is the normal scientific practice.²⁹²

Once obtained, database protection nominally expires after fifteen years.²⁹³ However, if the compilers make another substantial investment,say, by adding or updating new data to the preexisting collection, their efforts will renew the protection of the entire database for another fifteen-year period.²⁹⁴

In this respect, the sui generis database protection laws paradoxically provide stronger protection for derivative compilations than for derivative works obtained under traditional copyright laws. In the latter case, an original and creative derivative work receives copyright protection only for the eligible new matter added to the preexisting matter.²⁹⁵ In the case of the data protection laws, where no originality at all is required for eligibility, any qualifying additional investment may renew the protection of the collection as a whole.²⁹⁶ Perpetual protection thus becomes an attainable goal for the first time in the history of intellectual property laws (disregarding, of course, trademark laws, which operate on fundamentally different principles).²⁹⁷

In a series of recent decisions, the European Court of Justice (ECJ) has subsequently introduced an elusive subject-matter distinction between substantial investment for purposes of obtaining data that are created ( presumably ineligible), and expenditures for purposes of obtaining data that are collected (i.e., developed and maintained in databases as such) and which presumably qualify for protection.²⁹⁸ In other words, “only resources used to collect data that [are] already in existence” will qualify for database protection, but not “data compilations that are generated quasi ‘automatically’ as by-products of other activities.”²⁹⁹ To the extent that scientific databases are characterized as “created” under this slippery distinction, it might conceivably reduce the total number of databases, particularly sole-source databases, eligible for protection.³⁰⁰ Courts could, for example, exclude some collections of raw scientific data on these grounds.³⁰¹

However, some commentators believe most scientific data are better characterized as collected and, therefore, automatically eligible for protection.³⁰² Even when scientific data are viewed as created, whatever this turns out to mean, entities seeking protection could always spend more money on verification or on improving the conditions of access to and posterior maintenance of the collection, which might have some scientific value even if undertaken for secondary motives. In other words, there is reason to believe that most collections of scientific data and information could be made to fit within these judicially contrived eligibility requirements by one means or another. If so, any collection of scientific data or information that did qualify would obtain broad and virtually endless protection against value-adding components of a future collection that made unauthorized use of an existing one.³⁰³

How the Database Directive actually affects science in any given country will then depend on a number of uncertain variables. In the United States, where the scientific community vigorously opposed enactment of database protection bills modeled on the European Union Directive,³⁰⁴ only copyright law applies to compilations of data, although that law, as shown earlier, is much less science friendly today than in the past.³⁰⁵ In European Union member states and affiliates, however, the sui generis database protection laws remain firmly in place despite serious criticism from within the European Union itself.³⁰⁶ The European Commission has also made strenuous efforts to extend similar database regimes to developing and Least-Developed Countries through a series of regional and bilateral free trade agreements.³⁰⁷

The sui generis database protection laws in the European Union thus turn the relatively benign approach of traditional copyright law upside down. They not only protect the very aggregates of facts and data that international copyright law expressly left in the public domain, they confer potentially stronger and longer protection on these unoriginal compilations than copyright laws afford original works of authorship.³⁰⁸

Analogies drawn from the historical rhetoric promoting authors’ rights, whatever one’s view of them, were thus perversely applied to an investment-based scheme of protection governing the most fundamental building blocks of knowledge.³⁰⁹ What the sui generis database laws actually codified instead was a scheme of powerful exclusive property rights that protect infinitely expansible collections of data from extraction and reuse, with a built-in propensity to favor the emergence of sole-source providers over time.³¹⁰ This regime conflicts head on with customary scientific research practices that long antedated the digital universe and the accelerated research opportunities it makes possible.³¹¹

Nor should one suppose that the social costs of this dismal experiment, which cannot be repealed despite sweeping criticism from the Commission’s own officially appointed reviewers, are confined to the some fifty-five countries that have adopted similar regimes at the behest of the European Communities.³¹² Consider, instead, that because science is a global public good,³¹³ search engines and other digitally empowered research tools must transcend national borders in order to access all publicly available sources of data and information relevant to any given project. Standing in their way are all the formidable legal barriers rooted in the territorial copyright and database protection laws described above, which threaten to choke the transnational flow of upstream scientific data and information that would otherwise be capable of digital integration on a global scale.³¹⁴

II. EMPOWERING DIGITALLY INTEGRATED SCIENTIFIC RESEARCH ON A GLOBAL SCALE

The foregoing analysis of the existing intellectual property framework portrays a set of rules and policies that are diametrically opposed to the needs of scientific researchers in a universe of discourse where automated knowledge discovery tools must freely explore the entire range of thematically relevant, digitally distributed literature and data.³¹⁵ Consider, for example, that the Wellcome Trust found that eighty-seven percent of the material housed in the United Kingdom’s main medical research database (UK Pub Med Central) was unavailable for legal text and data mining.³¹⁶

By the same token, a major independent study undertaken for the British Government reports that existing copyright laws make it virtually impossible to text mine about one thousand journal articles from the first half of the twentieth century that describe malaria in indigenous peoples and soldiers, as well as details of therapeutic measures available at that period.³¹⁷ Because of rights clearing requirements that appear out of all proportion to any benefit the rights holders could want, “even if they could be found,” researchers cannot digitally index or text mine sources that offer potentially significant insights for the development of methods for preventing and treating malaria today.³¹⁸

A. Automated Knowledge Discovery Tools as Instruments of Massive Infringement

Wittingly or unwittingly, these laws force scientific researchers to choose between ignoring an unmanageable and unreasonable set of legal constraints, in the interest of pursuing science as a public good, or foregoing research opportunities in order to avoid thickets of rights, burdensome transaction costs, and the fear of stirring up potential law suits down the line. The end result puts both science and the larger public interest in a no-win situation, at a time when the resources available to fund scientific research are shrinking.

If the relevant intellectual property laws were strictly enforced, and the scientific community continued to respect them, scarce public resources earmarked for basic research would be siphoned off to intermediaries from scientists seeking access to and use of their own published research results. In that event, the public pays twice for the same output, plus a surcharge for mushrooming transaction costs, while the “incipient transnational system of innovation,” established by the TRIPS Agreement in 1996,³¹⁹ is progressively deprived of essential knowledge assets. Less innovation, not more, is the predictable result over time.

Conversely, if intellectual property laws are ignored by researchers determined to carry on with their work irrespective of unreasonable legal constraints, automated knowledge discovery tools will have become transformed into engines of massive infringement.³²⁰ It is hard to see how systematic disregard of intellectual property laws, coupled with growing contempt for the legislative process that fosters them,³²¹ will benefit authors, artists, and other creators in the long run, especially when those condemned to outlaw status are not free-riders on costly musical and cinematic productions, but publicly funded scientific researchers in pursuit of greater knowledge and applications that benefit humanity as a whole.

While the pressing need to reform the laws that have produced such anomalous results has not escaped notice,³²² efforts in this regard are confronted with a conflict between the interests of scientists, on the one hand, and those of publishers on the other. Scientists are authors whose primary interests in publication are the rewards of attribution and integrity—the so-called reputation benefits—that the moral rights of copyright laws, together with the norms of science itself, strive to protect.³²³ These reputational benefits then serve to attract the kind of financing and status rewards attendant on academic success.³²⁴ Given a conflict between the needs of scientific research and the dictates of copyright and database laws, one can expect scientists normally to opt for the goals of research because their pecuniary interests lie elsewhere, and are, indeed, dependent upon the reputation benefits just described.³²⁵

In contrast, science publishers are the main pecuniary beneficiaries of the current state of the law, which they have lobbied hard to obtain, and they would resist any reforms likely to be put on the table.³²⁶ This fact of life makes it logical to ask why the scientific community continues to rely and depend on publishing intermediaries in the first place. Disregarding the historical origins of such reliance, one feels compelled to ask whether the benefits of such reliance still outweigh the costs in today’s digitally integrated, totally computerized research environment. No sensible scheme of reform can be devised without addressing these questions, and no specific proposals will make sense unless they are weighed against alternative options that result from such an enquiry.

B. The Limits of Incremental Legislative Reform to Alleviate Obstacles to Scientific Research

To the extent that publishers retain their traditional role as intermediaries, any efforts to reform applicable intellectual property laws must reconcile the needs of science with the needs of commercial publishers to turn a profit.³²⁷ This factor greatly complicates the prospects for reform because the existing copyright and database laws so favor the interests of publishers over those of scientists that merely incremental or piecemeal reforms rooted in traditional exceptions and limitations are unlikely to give the research community what it needs.

A more promising approach might emerge if the scientific and publishing communities were to negotiate an industry-wide settlement that accommodated the research needs of one without sacrificing the commercial needs of the other. Arguably, one such example was the negotiated settlement between publishers and the library community with regard to photocopying, which replaced the case-by-case fair use approach in the United States with § 108 of the 1976 Copyright Act.³²⁸ However, librarians have found this settlement unsuited to the digital age, and they are demanding widespread reforms requiring legislative enactments of limitations and exceptions that publishers strongly oppose.³²⁹ The extraordinary powers that publishers have obtained under the DMCA in the United States and the InfoSoc Directive in the European Union make an industry-wide settlement favorable to science far more difficult now than it might have been prior to the 1990s.

In the remainder of this Article, we discuss possible solutions to the problems that intellectual property laws have created for digitally integrated scientific research from two very different angles. First, we consider the kinds of legal reforms that would be needed if commercial publishers continued to act as intermediaries between producers and users of scientific information and data, as they do today, without regard to the likelihood that such reforms would ever be enacted.

We then reconsider the role of publishers as such and ask whether, from a cost-benefit perspective, it should be significantly modified or abandoned altogether. In that Section, we examine alternative strategies that the scientific community itself could embrace in a concerted effort to manage its own upstream knowledge assets in ways that might avoid, or at least attenuate, the obstacles to digitally empowered scientific research currently flowing from a flawed intellectual property regime.

1. Possible Reforms of Domestic Copyright Laws

To the extent that the fruits of basic scientific research continue to stimulate economic growth in advanced industrial economies,³³⁰ improving the environment for digitally integrated scientific research will enhance the prospects for future innovation. In the words of an authoritative report to the British Prime Minister in 2011:

Innovation may be blocked and growth hampered when unduly rigid applications of copyright law enables rights holders to block potentially important new technologies…. Research scientists, including medical researchers, are today being hampered from using computerized search and analysis techniques on data and text because copyright law can forbid or restrict such usage…. In these circumstances, copyright in its current form represents a barrier to innovation and economic opportunity.³³¹

This, along with other studies recently undertaken, could eventually lead to proposals for incremental legislative reforms that would move in the right direction.

For example, officials of the European Commission recently undertook an enquiry into the ways that limitationsand exceptions in copyright laws might be improved with specific regard to scientific research.³³² As a result, the European Union might conceivably decide to revise the InfoSoc Directive. Alternatively, it might wait for an overall codification of European Union copyright law, as contemplated in the Treaty on the Functioning of the European Union (FEU).³³³

In the United States, a group of scholars has been considering the need to update existing limitations and exceptions with a view to a prospective revision of the 1976 Copyright Act.³³⁴ The National Academies of Science has just commissioned a study of the impact of copyright laws on scientific research.³³⁵ A thorough-going study of this same topic as it affects developing countries is also under way within the ambit of the WIPO Development Agenda.³³⁶

These and other similar initiatives might conceivably restore a better balance between public and private interests than currently exists under the global copyright regime as strengthened since the 1990s. To this end, the next Section of this Article outlines a set of incremental reforms that could at least attenuate the obstacles to digital research that were identified above.

However, we remain skeptical that proposals for incremental reform, even in the unlikely event of legislative enactment, would adequately address the roots of the problem. As we view the matter, the head on conflict between e-science and copyright law depicted above cannot be resolved without fundamental legal and institutional reforms designed to prevent both copyright and database protection laws from reaching into the domain of basic scientific research in the first instance.

a. Improving the Information Society Directive in the European Union

Any serious reform effort in the European Union should start with a codification of the idea-expression principle, a subject-matter exclusion of fundamental importance for scientific research.³³⁷ Most scientific literature conveys ideas and facts, not expression. Although the TRIPS Agreement and the WCT both embodied this exclusion in international copyright law,³³⁸ the drafters of the InfoSoc Directive conveniently ignored it in 2001,³³⁹ perhaps on the technical groundthat it was not a limitation on, or exception to, authors’ exclusive rights as such. All European Union member states should have to embody this principle in their domestic copyright laws.

Turning to the express exception for scientific research that the InfoSoc Directive introduced in Article 5(3)(a),³⁴⁰ the first step in any incremental set of reforms would be to make this exception also mandatory and binding on all member states.³⁴¹ Absent such a measure, countries such as the United Kingdom might simply ignore this provision and continue to rely on older exceptions allowing quotations for certain purposes,³⁴² copies made for libraries and educational establishments,³⁴³ as well as the new provision subjecting “private use” to fair compensation.³⁴⁴ The byzantine snares emanating from domestic law implementations of such narrow provisions are exemplified in a recent survey of the relevant U.K. laws.³⁴⁵

Second, the express exception for science in Article 5(3)(a), once made mandatory, must be rid of its inherent ambiguity.³⁴⁶ Like the relatively more transparent phrase used in the Rome Convention of 1961, for example, the revised provision could directly permit “use solely for the purposes of teaching or scientific research”³⁴⁷ and thus remove any reference to the confusing term for “purpose[s] of illustration.”³⁴⁸

Third, any mandatory exception for scientific research must then be cloaked in some substantive content that promotes flexibility within an inherently proscience framework and deflects narrowing legalistic interpretation in advance. In particular, the revised provision should eliminate the current language that limits scientific use “to the extent justified by the non-commercial purpose to be achieved.”³⁴⁹ This language is unworkable in practice because, as we have noted earlier, virtually all scientific research conducted at today’s universities and other public research entities can be perceived as abetting commercial ends that financially benefit their sponsors.³⁵⁰

To achieve even these minimalist proscience ends without departing from existing legislative models, the European Commission should also consider embedding a mandatory exception for science within a broader fair use framework, like that adopted in § 107 of the U.S. Copyright Act of 1976.³⁵¹ Recent European scholarship endorses this approach,³⁵² although opposition to it is strongly entrenched in business circles.³⁵³ However, United States fair use law retains potential defects of its own that could limit its effectiveness if used to regulate digital research in the European context. Accordingly, suggestions to improve the fair use approach in the United States highlighted below should also be considered for possible application in the European Union context.

b. Improving the Fair Use Approach

We previously pointed out that the fair use approach in the United States, while more flexible than the designated exceptions approach in the European Union, could not readily cope with either quantitative or qualitative amounts of copyrighted matter that digitally driven scientific research would have to process. Similarly, application of the market-harm test in such cases might be difficult if publishers successfully insisted that such uses constituted the natural market for their proprietary outputs.³⁵⁴

Here, we note a further possible snag in the transformative use doctrine described earlier, which the federal appellate courts have recently expanded as a tool for equating public good uses of protected works with presumptively fair uses.³⁵⁵ Tensions arise because the very concept of transformative use partakes of the definition of a derivative work,³⁵⁶ and U.S. copyright law gives strong protection to derivative works.³⁵⁷ Today, indeed, some U.S. courts have begun to distinguish transformative markets from transformative uses, which captures the exquisite ambiguity of the underlying concept and could begin to wrap so-called transformative uses in the numbing foil of market-failure analysis once again.³⁵⁸

In the leading Supreme Court decision on fair use, Justice Souter dropped a footnote identifying this very conflict.³⁵⁹ He suggested that a judicially imposed license allowing a transformative use with equitable compensation to the derivative right holder could resolve the dilemma in close cases.³⁶⁰ To date, no U.S. court has taken the hint, which is why U.S. fair use decisions often vacillate between all-or-nothing outcomes in a path that sometimes defies logic or rationalization.³⁶¹ Perhaps the recent pertinent decision by the U.S. Supreme Court in eBay, Inc. v. MercExchange, LLC,³⁶² will finally focus the copyright courts’ attention on the possibility of using a liability rule, in place of an injunction, in appropriate cases.³⁶³

On this issue, the three-step test familiar from European law (and now mandatory under international law) may have a valuable lesson to teach U.S. courts. In cases where normative considerations sounding in the larger public interest favor a given use, but the amount of the taking appears relatively large with some negative impact on the potential market for the copyrighted work, the legislative history of the three-step test would support allowing that use in return for equitable compensation from the proceeds of the otherwise unauthorized use, if any, to the authors whose support of the public interest had thus been co-opted.³⁶⁴ Should the European Union decide to adopt a modified fair use provision along these lines,³⁶⁵ it might move world copyright law toward some new synthesis that could combine the normative wisdom of U.S. fair use law with the practical wisdom of those reticent drafters of the gloss on Article 9(2) of the Berne Convention.³⁶⁶

The trouble even with this sort of adjustment is that it would probably not meet the needs of twenty-first century computational science, however beneficial it might be in other areas of literary and artistic endeavor. Because U.S. fair use cases remain so fact specific, the four normative criteria set out in § 107 of the Copyright Act could play out differently when tested before different judicial panels. In particular, the amount of material taken for digital research and, increasingly, included in new research results could always make some courts fearful of undermining the derivative work right, as mentioned above, even though strong derivative work rights make economic sense only in the entertainment sector.

Much would depend on the federal courts’ continued willingness to defend the transformative uses of science in the name of an overriding public interest. Even then, some decisions—though often criticized—introduce into U.S. fair use law the same untenable distinction between so-called commercial and noncommercial scientific research³⁶⁷ that European Union law has codified in its basic exception to the reproduction right favoring science.³⁶⁸ Because we believe that U.S. fair use law will have to take the internationally mandated three-step test more fully into account as time goes on³⁶⁹ (at least where foreign authors’ rights are at stake),³⁷⁰ this element alone could add an additional reason to fear a chilling effect on scientific research stemming from the uncertain application of the fair use doctrine to digital and computational science.³⁷¹

To obviate this uncertainty in U.S. law, economist Paul David has proposed codifying an “automatic fair use exception” for these purposes.³⁷² That exception could operate in tandem with voluntary contractual waivers, like those of the Creative Commons and Science Commons initiatives,³⁷³ discussed below. By the same token, the Hargreaves Review favors a new exception in the United Kingdom’s copyright law allowing uses enabled by technology that do not directly trade on the underlying creative and expressive purpose of the work.³⁷⁴ Reportedly, the U.K. government is favorable to this proposal.³⁷⁵ Some clearinghouse arrangements might nonetheless become necessary for purposes of guaranteeing reputational benefits through proper attribution.³⁷⁶

A codified automatic fair use provision for e-science, or at least a strong normative guideline to the same effect, would not impede the publishers’ ability to price discriminate their initial subscriptions in keeping with the subscribers’ capacities to pay.³⁷⁷ Both print publishers (whose numbers are decreasing)³⁷⁸ and online publishers (discussed below)³⁷⁹ could legitimately extract more revenue from commercial entities than from public science institutes under this approach. An automatic fair use provision might also further encourage commercial publishers to accept open access subsidies from science funders, a trend we discuss below.³⁸⁰

However, even an enlightened fair use approach raises obstacles rooted in print media models that ought to simply become irrelevant to the conduct of online, worldwide scientific research. In this context, for example, it makes little sense to focus on “reasonable” uses of published scientific articles,³⁸¹ or to attempt to track revenue streams from upstream uses of published scientific information and data by researchers who exploit automated knowledge discovery tools.³⁸² On the contrary, we believe intellectual property laws should not permit publishers to further control uses or reuses of their authors’ scientific research results for purposes of further research at all.

2. What E-Science Really Needs from Any Legislative Reform

We doubt that the foregoing proposals to incrementally reform existing measures bearing on scientific research could be enacted in an uncompromising format that would provide digital science with the user-friendly regime it needs to flourish. Any such proposals could easily become entangled in the coils of more intricate, legalistic provisions largely derived from experience in the entertainment sectors. Precisely because these so-called reforms would be deemed science friendly in name, they could mire modern science ever more deeply in the need to make unpalatable choices between obeying complex, inherently obsolete provisions or ignoring them altogether.

a. A Tailor-Made Exemption for Scientific Research

The only workable solution is to adopt a broad and uncompromising exemption for scientific uses that requires no gloss, no fine print, and no elaborately contrived exceptions to a grudgingly acknowledged “exception” for scientific research. To this end, the Max Planck Institute’s response to the European Commission’s Green Paper in 2008³⁸³ proposed that such a broad and general provision, allowing use and reuse of published research materials for virtually any scientific purpose, should expressly legitimize storage, archiving, data extraction, linking, and the like.³⁸⁴

While endorsing this proposal, which makes a good start, we think even more may be needed. In particular, scientists must be free to subject any published article (and, as we shall see later, any article made publicly available online)³⁸⁵ to data mining procedures and data manipulation by automated knowledge discovery tools, including virtual scientific experimentation, without any constraint other than attribution under the norms of science.³⁸⁶ The same exemption must apply to the public release of selectively chosen material in any scientific paper or report. Such a regime should be applied directly, and in harmonized express terms, in the copyright laws of every European Union member state, without any allowance for the sort of off-setting, detailed provisions that are currently thought necessary for “a workable system … of users’ rights,”³⁸⁷ which in practice usually means an unreasonable system of publishers’ constraints on science.

Such a broad exemption should expressly clarify its application to so-called derivative works, a concept that has virtually no meaning in upstream scientific research as currently practiced. So long as prior research results are incorporated into new scientific work with clear and appropriate attribution, there is no need for permission, which, in effect, operates as a de facto prior restraint on scientific speech.³⁸⁸ Nor should any commercial/noncommercial distinction be embedded in the copyright law’s broad research exemption, for the primary reason stated earlier, that basic scientific research results are properly to be treated as a public, not a private good, regardless of their origin.³⁸⁹

Effort should then be made to persuade the United States and other WTO members to adopt a similar provision, over and above any existing fair use provisions. Pending the formulation of a truly transnational science funding entity in the European Union, the Commission’s own science funding division should help to enforce such a broad scientific research exemption, as well as any other proscience provisions that may exist in the domestic copyright laws. The developing countries should also throw their weight behind a strong exemption for scientific research, in keeping with the WIPO Development Agenda.³⁹⁰

b. Breaking the Digital Locks

No provision exempting scientific research from the exclusive rights of copyright law, as proposed above, could fully achieve its purpose unless complementary legislative action were taken to ensure its effectiveness in the online environment. Here we encounter the blocking effects of technical protection measures (TPMs) as implemented in the domestic laws,³⁹¹ whose drafters ignored the proscience mandate expressed in the preamble to the WCT itself,³⁹² as well as other balancing provisionsset out in that treaty.³⁹³

If rights holders who make scientific works available through digital networks can simply enclose those works behind technological fences and then abolish all user-friendly provisions by contract, little would be gained by clarifying the idea-expression dichotomy or the scope for private and fair uses, or by enacting broad exceptions for scientific research and teaching as advocated above. The imposition of private intellectual property rights by such technological means³⁹⁴ also raises profound conflicts with constitutional law in the United States³⁹⁵ and with fundamental rights in Europe.³⁹⁶

In effect, publishers of digitally transmitted scientific articles online have the same legal entitlements as owners of music, films, and other cultural assets under the framework established by the DMCA in the United States and by parallel legislation in the European Union.³⁹⁷ These provisions give publishers the best of two worlds. On the one hand, if Internet Service Providers (ISPs) transmit copyrighted scientific articles without permission, publishers may force them to remove the offending publications under a set of provisions known as the “notice and take down” regime.³⁹⁸

Scientific researchers, on the other hand, like users of cultural goods from the mundane to the sublime, enjoy virtually no analogous powers to oblige online publishers to respect any of the privileges and immunities that are supposed to defend their interests.³⁹⁹ On the contrary, copyright laws protecting digital transmissions give publishers virtually absolute control over use and dissemination, as reinforced by their ability to impose licensing terms and conditions backed up by the impenetrable electronic fencing discussed earlier.⁴⁰⁰

Science publishers, who thus combine private law tools with the exclusive rights of intellectual property laws, can operate as unregulated ISPs under no obligation to respect the sharing norms of science and with a direct financial interest in deciding whether or not collaborative research and innovation will occur, and under what terms and conditions. Yet, there is simply no policy justification for subsuming the needs of science, with its far different economic and social considerations, and far larger impact on human life, to the current undiscriminating framework of copyright and database protection laws.

Some courts in both jurisdictions have begun to push back against these controversial digital locks,⁴⁰¹ and numerous proposals have been made for legislative or administrative solutions to pry them open.⁴⁰² For example, some have suggested a system of “electronic locks and keys,” which, however, could trigger costly and burdensome administrative procedures that could indirectly exert a chilling effect on users’ freedom to build on preexisting scientific and technological data and information.⁴⁰³ Professor Dan Burk has proposed a doctrine of “anticircumvention misuse” to deal with this same problem,⁴⁰⁴ while Professors Reichman and Franklin would impose “a fair and reasonable terms,” standard on all non-negotiable restrictions on access to and uses of computerized information goods.⁴⁰⁵ Still other proposals, while not without merit, would generally entail a considerable amount of political and legislative momentum and, unless carefully implemented, could in some cases complicate rather than simply avoid existing obstacles.⁴⁰⁶

Legislatures enacting appropriate exceptions for scientific research, like the one proposed above, should also simultaneously implement the proviso set out in Article 11 of the WCT, which expressly exempts “acts … which are … permitted by law” from the obligation of signatories to “provide adequate legal protection and effective legal remedies against the circumvention of effective technological measures.”⁴⁰⁷ For example, the copyright revision bill now languishing in Brazil initially took a major step forward by prohibiting content providers from using TPMs to defeat privileged uses or to impede access to public domain matter.⁴⁰⁸ Whether these and other provisions that seek to expand the copyright misuse doctrine⁴⁰⁹ will survive the legislative process in that country remains to be seen, as are the means of implementing them in practice, which future regulations would have to specify.

Meanwhile, one relatively expedient suggestion is the “reverse notice and takedown” regime put forward by Professors Reichman, Dinwoodie, and Samuelson.⁴¹⁰ Under their proposal, bona fide public interest users could avoid passing through a content provider’s electronic gateway and, instead, send a request or “flaming arrow” over the electronic fence to catch the copyright proprietors’ attention.⁴¹¹ This notice would signal that the user intended to obtain specified matter held by the proprietor in an online repository for purposes allowed under specified limitations and exceptions.⁴¹² It would give proprietors a period—say fourteen days—in which to accede to the request or deny it on specified grounds that it was willing to defend in court or an administrative proceeding.⁴¹³

In the latter event, both sides would know that a judicial test of the validity of the request under relevant exceptions would be the likely outcome, and the copyright authorities could establish an expedited judicial or administrative procedure for this purpose.⁴¹⁴ Once the legitimacy of the request was established, the relevant authority or court could enable third parties, if necessary, to disarm or decrypt the TPMs in order to extract the desired scientific material for the specified research purposes.⁴¹⁵ Publishers who needlessly barred the initial request and thereby necessitated a judicial inquiry should bear at least the transaction costs and might be made subject to additional penalties for abuse of TPMs.⁴¹⁶

While a “reverse notice and takedown” regime might entail palpable transaction costs at the outset, it would likely give rise to a jurisprudence of exceptions to TPMs that would, over time, facilitate use of the method.⁴¹⁷ Besides, to the extent that a broad exemption for scientific research purposes were enacted along the lines indicated above, requests for access to and use of technically protected data and information should normally elicit an automatic positive response.⁴¹⁸

Recent case law in the United States has made judicial resort to a reverse notice and takedown procedure more feasible even without enabling legislation.⁴¹⁹ However, given the massive amounts of literature and data processed by automated knowledge discovery tools, even the reverse notice and takedown regime—backed by supporting judicial decisions—could break down unless published scientific works in general were governed by some globally effective “digital copyright exchange,” like that recommended in the Hargreaves Review.⁴²⁰ Even then, much would depend on the willingness of science funding agencies to insist that science publishers either refrained from surrounding scientific works transmitted online with TPMs and DRMs or that they made such works automatically accessible to scientists seeking access to them through approved portals for research purposes.

In sum, absent some procedure like the reverse notice and take down regime for freeing up unprotectable scientific information, the TPMs become a means of inducing massive abuses of the copyright law,⁴²¹ much as peer-to-peer file sharing can become an instrument for inducing massive infringements of exclusive rights.⁴²² A reverse notice and take down regime would at least enable scientific researchers to avoid access controls and any resulting electronic contracts that imposed waivers of statutory limitations and exceptions or other harsh restrictions on use and reuse of privileged information and data. This feature should make it particularly attractive to the European Commission in that it would finally provide them with a practical means of fulfilling the obligation that Article 6(4) of the InfoSoc Directive already imposes on member states to ensure the availability of the specified exceptions set out in Article 5 when implementing the Directive itself.⁴²³

c. Disciplining Contractual Overrides

The foregoing discussion demonstrates that no set of limitations and exceptions enacted by enlightened legislators can achieve the goal of strengthening scientific research so long as the proprietors of scientific publications can contractually override them, whether in print media or in the online environment. For this reason, the Max Planck Institute rightly proposes that both new and existing exceptions favoring scientific research must be made peremptory, mandatory, and nonwaivable.⁴²⁴

Short of this logical proposal, other important, if less efficacious measures, remain available. For example, Professor Burk’s principle of anticircumvention misuse mentioned earlier could be adopted on both sides of the Atlantic to limit private interference with specified public good uses of copyrighted works.⁴²⁵ To the same end, Professors Reichman and Franklin’s proposals for a “public interest unconscionability” standard for non-negotiable contracts could be employed to give courts more common law tools for alleviating conflicts between private ordering and the goals of federal copyright and related laws.⁴²⁶

There is reason to believe such a tool would fit well within certain existing European approaches to consumer protection and contract laws in general.⁴²⁷ Professor Hilty also stresses the possibility of invoking European competition law, with its concept of abuse of a dominant position, when proprietors leverage their power in the market for scientific articles to inhibit use and reuse of scientific contents by downstream investigators.⁴²⁸

What matters is that legislatures concerned with promoting scientific research should take a forthright position against contractual overrides of lawful and permitted uses while also clarifying scientific research as a peremptory example of a lawful and permitted use. In reality, however, there is no reason to expect any such enlightened approach in the immediate future. On the contrary, newly proposed measures on enforcement, in their present form,⁴²⁹ could actually strengthen the proprietors’ ability to impose privately legislated intellectual property rights⁴³⁰ on the scientific research community.

c. Aligning Database Protection Laws with Broad Exceptions for Science in Copyright Law

Any legislative reform of domestic copyright laws that ignored the database protection laws in the European Union would inadvertently allow the latter to surround the former with a net that would block access to and use of the very facts and data that the copyright paradigm ostensibly left free.⁴³¹ It would also impede transnational efforts to pool large collections of scientific data by automatically subjecting contributions from providers in the European Union to a strong regime of exclusive property rights not applicable to other contributors.⁴³² For these and other reasons, neither science nor culture⁴³³ could fully attain the payoffs that digital technologies make possible without ancillary adjustments of the Database Directive.

When the Max Planck Institute called for a broad exemption from the exclusive rights of the European Union’s domestic copyright laws for published scientific information and data, it logically demanded that the Commission should also insert similar language into the Database Directive as well.⁴³⁴ In effectuating any such alignment, the Institute insists that the exceptions for science in both copyright laws and database protection laws should be preemptory, mandatory, and immune from both contractual overrides and TPMs.⁴³⁵

As was the case with copyright law, a broad exemption that clearly allowed extraction and reutilization of non-copyrightable data for scientific research must expressly empower the use of automated knowledge discovery tools for this same purpose.⁴³⁶ Such language should ensure the rights of scientists to aggregate data and information in a research commons, to conduct data mining and similar techniques, and to extract data embedded in scientific articles for use in further research.⁴³⁷

To the extent that the production of scientific data remains largely government-funded, no exclusive property rights should normally attach, even to downstream commercial uses of such data.⁴³⁸ However, commercial uses of large semi-autonomous data sets as functional tools, such as microarrays or diagnostic tools,⁴³⁹ could become suitable candidates for equitable compensation under nonexclusive licenses.⁴⁴⁰ In such cases, a compensatory liability regime—i.e., a take-and-pay rule described earlier—might provide a workable incentive without the blocking effects that patents tend to impose on research tools in general.⁴⁴¹

Still other measures are necessary to attenuate the deleterious effects that the European Commission’s Database Directive has imposed on all scientific, educational, and cultural pursuits that depend on ready access to published facts, data, and information. For example, compulsory licenses should become available when the database is the sole source for the data in question.⁴⁴² The Directive as approved by the Council of Ministers was stripped of such a provision at the last minute, and the importance of restoring a comparable provision is clear from hindsight.⁴⁴³ The potentially unlimited duration of database protection also remains an untenable assault on basic principles of intellectual property law. Provision for the entrance of older data into the public domain after a specified period of expiry should be a governmental priority even as new data added to the collection attracts new protection rights.⁴⁴⁴

C. Adjusting the International Legal Framework to Accommodate the Needs of Science

The prevailing international minimum standards of intellectual property protection are not necessarily in conflict with the proposals set out above. First, the standards themselves are broad and open to interpretation, as will be shown in more detail below, while both Article 1.1 of the TRIPS Agreement and Article 14(1) of the WCT contain crucial deference provisions that deliberately leave room to maneuver when states make a good faith effort to conform these standards to national needs and policy.⁴⁴⁵ Second, the flexibility built into the TRIPS and WCT standards applies in two directions. Although tightening the exclusive rights with more restrictive conditions is always an option,⁴⁴⁶ it remains equally possible to flesh out the limitations and exceptions, along with other balancing features, in a manner more favorable to the provision of public goods than has been the case in some OECD countries and in many developing countries as well.⁴⁴⁷

For these and other reasons, we remain confident that the positive law mandates of the treaties do not negate the proposals for reform outlined above, so much as a lack of political will and an absence of the kind of collective action needed to stimulate it. In what follows, nonetheless, we devote particular attention to the three-step test itself, which some consider the biggest obstacle of all to reform.

1. Reinterpreting the Three-Step Test

At least one expert believes that the three-step test already allows more open-ended assessments of both existing and future limitations and exceptions, in the manner of U.S.-style fair use decisions, than many courts and commentators suppose.⁴⁴⁸ On this view, the extension of the three-step test to all of copyright law would actually provide a tool—if properly worked—that could help to deal with fact-specific cases, without necessarily undermining the force of general exceptions for research and education.⁴⁴⁹ Support for this view exists in a number of recent decisions by German Courts,⁴⁵⁰ in the Agreed Statement to Article 10 of the WCT,⁴⁵¹ and in the willingness of one WTO panel to read the TRIPS Agreement in light of subsequent developments under Article 10 of the WCT itself.⁴⁵² Unfortunately, the EC’s InfoSoc Directive ignored these openings and deliberately used the three-step test to further confine even preexisting limitations and exceptions in the copyright laws of member states,⁴⁵³ an outcome Professor Senftleben has deemed “a worst case scenario.”⁴⁵⁴

More promising in this regard are recent proposals from the Max Planck Institute for judges applying the three-step test, which could induce them to undertake a more normative analysis than in the past.⁴⁵⁵ That type of analysis is something European positivist courts are unaccustomed to doing,⁴⁵⁶ although under a fair use provision, as codified in U.S. copyright law in 1976, for example, courts must routinely perform this very task.⁴⁵⁷

The Max Planck proposals deliberately build on the preamble to the WCT, which recognizes “the need to maintain a balance between the rights of authors and the larger public interest, particularly education, research, and access to information ….”⁴⁵⁸ In that vein, the proposal would:

• Mandate that courts applying the three-step test falling under Article 13 of the TRIPS Agreement in copyright cases take into account the interests of third parties, including individual and collective interests of the general public, and not just the interests of rights owners;⁴⁵⁹

• Avoid prioritizing any one step, or requiring an affirmative answer to all steps, but would instead require a judicial balancing of the different prongs, as occurs under U.S. fair use law;⁴⁶⁰

• Give particular weight to unauthorized uses that are underpinned by fundamental rights⁴⁶¹ and other “common interests,” notably “in scientific progress and cultural or economic development;”⁴⁶²

• Seek to promote competition, especially in secondary markets, by a correct balancing of interests, but without making the three-step test a proxy for competition law;

• Expressly recognize that adequate compensation may be less than market pricing where other public concerns are at stake, includingthird-party interests or the general public interest.⁴⁶³

The Max Planck Institute’s carefully considered reforms would introduce a healthy dose of legal realism into the traditional positivism surrounding European copyright jurisprudence. They would counter the prevailing notion in Continental copyright law, which favors narrowly confined exceptions in deference to the authorial interest.⁴⁶⁴ They would also curb the European Commission’s tendency to fall back upon a market failure rationale for limitations and exceptions,⁴⁶⁵ a tendency from which U.S. courts have increasingly retreated in recent important decisions bearing on fair use.⁴⁶⁶

2. Leveraging the WIPO Development Agenda

The outlook for these proposals could depend in part on the continued spread of the fair use doctrine beyond the United States,⁴⁶⁷ and on the extent to which the developing countries affirmatively responded to them within the ambit of the WIPO Development Agenda.⁴⁶⁸ This Agenda has already mandated formal scrutiny of limitations and exceptions under the prevailing copyright conventions, with a view to clarifying the extent to which they insufficiently promote access to knowledge in developing countries. The WIPO Development Agenda has also spawned a major normative reexamination of limitations and exceptions, prepared by Professors Hugenholtz and Okediji, to this same end.⁴⁶⁹

If influential WIPO members lent it their support, these initiatives could at least produce a soft-law declaration of normative content that might turn the three-step test into a pathway towards a proper users’ rights formulation.⁴⁷⁰ A soft-law instrument adopted at WIPO could become particularly influential if it endorsed or incorporated the Max Planck Institute’s own “Declaration on the Three-Step Test.”⁴⁷¹ If, moreover, a regional group of, say, Latin American, Asian, or African countries decided to implement proposals emerging from these deliberations in their domestic laws,⁴⁷² as Brazil had begun to do at the time of writing,⁴⁷³ it could trigger a broader movement forcodification of users’ rights at the international level.⁴⁷⁴

III. ENABLING E-SCIENCE TO MANAGE ITS OWN UPSTREAM RESEARCH ASSETS

So far, our focus on measures to make copyright and related laws more science friendly has operated on the premise that publishers would continue to play their traditional role in the process of disseminating research results. This very premise, however, makes it unlikely that the legislative or judicial reforms outlined above are implementable within the OECD countries in the near future, despite growing attention to the conflict between intellectual property laws and the needs of science in a digital age.⁴⁷⁵

The lobbying power of publishers has never been greater. Concerns about protecting the interests of the entertainment and cultural industries continue to elicit stronger intellectual property laws at both the national and international levels, with little or no regard for their potentially deleterious effects on scientific research or the provision of other public goods.⁴⁷⁶ Whether reform efforts underway in some emerging economies may create a countervailing trend is impossible to predict,⁴⁷⁷ but the benefits of such a trend—if it emerges—would likely play out over a lengthy period, and might not extend, at best, beyond certain regional alignments.

Science policy will, accordingly, have to evolve defensive measures of its own in order to neutralize interference from the default rules of copyright, contract, and database protection laws as they stand. Scientists, in short, will increasingly have to manage their own upstream research assets as global public goods, sheltering them within a reinvigorated sharing ethos, in the interests of a more productive downstream innovation system otherwise driven by the incentives of industrial property laws.⁴⁷⁸

As will be seen below, the scientific community, led by many dedicated and visionary individuals and institutions, has already taken steps to widen the choice of open distribution outlets for scientific literature and data. These promising initiatives nonetheless remain hampered by the community’s continued reliance on publishing intermediaries. In this Part, we first reevaluate the role that these intermediaries should play under existing institutional constraints. We then ask if better solutions are not likely to emerge from a change of paradigm, in which the outsourced intermediaries are either downgraded or abandoned altogether, open access modes of dissemination were to take their place and the knowledge production and scholarly communication functions were increasingly to be absorbed into digitally integrated thematic research environments.⁴⁷⁹

A. Reassessing the Role of Publishing Intermediaries

Until recently, the customary practice of the scientific community was to rely almost entirely on external publishing intermediaries, even though the bulk of published scientific research would have been government-funded.⁴⁸⁰ In conformity with this practice, authors of scientific articles normally assign their copyrights to publishers, who are either commercial entities or learned societies and other not-for-profit scientific organizations.⁴⁸¹ As a result, it was publishers, rather than authors, that initially determined the conditions for access to these same articles and for reuse of the information and data they contain.⁴⁸² At the same time, authors benefit from the peerreview mechanisms many of these publishers manage, which makes them reluctant to publish outside traditional, well-established or high impact outlets, when they have the choice.

1. Costs and Benefits of the Traditional Approach

Historically, the logic behind this custom was the need to defray high front-end publishing costs and to perform laborious tasks, such as typesetting and formatting, as well as the physical distribution of printed copies.⁴⁸³ A second factor was the willingness of many scientific subcommunities to entrust learned societies with the publication task, which in turn became a primary source of revenue for the societies whether they actually performed the publishing service, or, increasingly in recent years, outsourced it to a commercial publisher in return for a share of the proceeds. Over time, the possibilities for profit have enticed commercial publishers to buy out the learned societies, although some commercial publishers do make continuing payments.⁴⁸⁴

Lately, scholars have challenged such logic,⁴⁸⁵ and some have argued that the value added by such intermediaries has reached diminishing returns.⁴⁸⁶ The once costly front-end publishing function has increasingly been reduced to desktop publishing and automated formatting,⁴⁸⁷ while the peer-review function, of great importance to the integrity of science, is performed gratis by scientists who themselves gain power, reputation, and advanced access to new developments from their voluntary labor.⁴⁸⁸ This built-in quid pro quo within the scientific community has perpetuated the dominance of the proprietary intermediaries, along with the practice of negotiating the sale (now licensing) of subscriptions directly to libraries without inputs from users. Meanwhile, the supervisory or editorial role of the learned societies, with some exceptions, has diminished over time, although the dependence of such societies on income from publishing seems ironically to have increased.⁴⁸⁹

This web of traditional practices and interests carries into the digital age, even though digital networks offer repeated opportunities to break with the limits of the print model and make whole new dimensions of publishing possible. What really changes in the online environment are not the basic principles of scientific collaboration,⁴⁹⁰ so much as the burdens and role of publishing intermediaries in the sciences, who increasingly may never publish a physical print copy at all.

This growing tendency to rely on online distribution in the sciences has undermined prior balancing effects of the first sale principle under traditional copyright law.⁴⁹¹ For example, there are fewer printed copies extractable from initial revenues and then freely redistributed, and the subscription price per journal may rise prohibitively.⁴⁹² Even when printed copy distribution continues, the role of publishing intermediaries’ in the online environment changes radically, as they add less value to the authors’ own research results⁴⁹³ and become online service providers whose primary contribution to authors is convenience.⁴⁹⁴

Notwithstanding these changed conditions, the rules of copyright law have simultaneously been extended to the digital environment, and the protections available have been greatly strengthened, as demonstrated earlier, in order to make the online environment safe for the transmission of printed text. Because scientific publishing has drifted along with this tide, the full possibilities of digitally manipulating research results for new scientific discoveries are hamstrung by the layers of protection inherited from these legal and institutional developments, and there is a pressing need to avoid the resulting harm to science.⁴⁹⁵

The open access movement is a major response to this challenge. Today, an ever-growing number of scientific journals are published online, on a fully or partially open access basis,⁴⁹⁶ although these are not yet always the most prestigious journals in their respective fields.⁴⁹⁷ To the extent that the learned societies themselves resist the drive for greater use of open access modalities, their dependence on royalty streams from commercial publishers for scholarly pursuits and other activities may explain their reluctance to change more than the economics of publishing itself.

While the outsourcing of scientific journals to commercial publishers may still make sense, despite an array of other options, there is a growing trend to subsidize the open access format, even in an otherwise commercial context, as part of the publicly funded research process. The funding agencies, foundations, and universities that support specific research projects may thus provide supplementary funds to pay the commercial publisher a set fee in lieu of royalties or other compensation.⁴⁹⁸ In such cases, the funders may—and increasingly will—set open access terms as the quid pro quo of the subsidy itself.⁴⁹⁹ Commercial publishers are increasingly disposed to allow this option, and science funders have begun aggressively to insist on it in some disciplines,⁵⁰⁰ although the sustainability of this approach obviously depends on the continued availability of financial resources for this purpose.

The point is that desktop-publishing techniques and online transmission have made it technically (if not culturally) feasible to redefine the role of existing intermediaries who benefit from research-hostile intellectual property laws and practices. By the same token, once publicly funded research results are made available to the scientific community, with due respect for attribution, it becomes logical to ask why scientists qua users should ever pay scientists qua authors, irrespective of what the default rules of copyright and database laws provide to the contrary.⁵⁰¹

2. Redefining the Role of Publishing Intermediaries under Current Institutional Constraints

Given the diminished costs incurred by today’s intermediaries in the online environment, and the shrinking amount of added value they actually contribute under modern conditions,⁵⁰² one must logically ask what entitlements they should be allowed to claim for secondary uses of published scientific research results in either the print media or the online environment, and how such claims should be implemented when recognized. At bottom, what science publishers provide in the online environment are measures to maintain quality assurance and control, marketing and distribution, plus certain technical services that the research community could provide for itself, yet typically does not in rich countries, perhaps because of inertia. The reputational benefits that are of primary importance for authors accrue from the peer-review function that is largely provided gratis by other reputable scientists. The intermediaries’ utility stems from maintaining and updating electronic collections, possibly also from electronic indexing of these collections, and possibly from the provision of other technical services needed to make embedded data and information available upon request.⁵⁰³

Although science publishers must necessarily charge for these services, funding agencies should, and increasingly do, ensure that government-funded research results remain freely available in public or private repositories, so that to defray these costs, users could perform the needed technical services on their own.⁵⁰⁴ Such a policy also serves to attenuate the problems of sole-source providers, who monopolize public science and can pose serious challenges for digitally integrated scientific research.⁵⁰⁵

Recognizing that publishers must charge for their technical services need not extend to endowing them with exclusive rights to downstream uses or reuses of the scientific product they make available. On the contrary, the proprietary restrictions that such rights enable intermediaries to impose in the name of authors’ rights, without any palpable authorial contribution, should be swept away as inconsistent with both the needs of science and the principles of sound exceptions to copyright and database laws as expounded above.

Because publishers of scientific journals depend, in the first instance, on contractual relationships with the learned societies (or other sponsors of academic journals), these regulatory adjustments can be achieved by contract, without need of legislation.⁵⁰⁶ For example, institutional mandates can restrict the transfer of copyrights in publicly funded research results and require that such results be made available in appropriate repositories. Until this transformation of the current publishing model occurs, however, funders of scientific research—whether government agencies, foundations, or academic institutions—should insist on open access publishing conditions as part of the grant-making process.⁵⁰⁷

So long as the subscription model is preserved, intermediaries whose services are deemed of value to the relevant scientific communities should thus be required to allow scientists to make any and all needed research uses of published scientific articles, including full digital empowerment for uses of automated knowledge discovery tools, computational tools, and the like, without need for express permission, in exchange for fixed compensatory charges built into the online subscription price. That price could be tiered to reflect the for-profit nature of a subscribing entity, but it should never be calculated on a payper-article basis or the like.

Intermediaries would then be recognized for what they are, i.e., information brokers, and permissible charges would be negotiated on a reasonableness basis, one for example resembling the virtual-market criteria proposed by the Max Planck Institute.⁵⁰⁸ In this context, however, we stress that the “market” for published scientific articles is already an artificial construct to begin with, built more on captured products than on verifiably value-adding services,⁵⁰⁹ and increasingly sustained by compromise embargo periods following publication before some articles are deposited in open access repositories.⁵¹⁰

Under such a contractually reconstructed regime,⁵¹¹ scientists should have a right to use digitally provided content for any research purpose, including both personal use and redistribution, subject to the above-mentioned negotiated-service charges to cover the costs of delivery and maintenance. Disputes over the reasonableness of costs should not bar access to the use of these resources, but would have to be settled offline by mediation, arbitration or, as a last resort, litigation.⁵¹² Resort to collection societies could then be avoided precisely because there would be no need to monitor actual use for payment purposes.

Only the actual costs of the intermediaries’ brokerage services would need to be taken into account, along with a negotiated surcharge for profit.⁵¹³ All parties should understand that outer limits on the aggregate online service charges necessarily follow from the fact that taxpayers largely support the entire enterprise; from the need to conserve scarce resources for scientific investigation; and from the implicit threat that, if intermediaries refuse to cooperate, the funders themselves could support alternative arrangements, like those discussed below, including some institutionally organized not-for-profit providers.⁵¹⁴

In fact, the movement to implement open access scholarly journals has rapidly expanded in the past decade, with over 7500 journals currently operating on this basis.⁵¹⁵ Under this approach, authors, research funders, and institutions (or some combination thereof) cover the costs of publication. Absent such an approach, care must be taken to avoid fostering sole-source monopolies over unsubstitutable scientific materials that can never realistically be regenerated or otherwise readily obtained from public repositories.⁵¹⁶

B. Funders’ Ability to Contractually Regulate Access, Use, and Reuse of Scientific Literature and Data

Implicit in the foregoing analysis is the premise that most published scientific research results, at least in OECD countries, will have been funded largely by governments or nonprofit foundations. These entities have the power to impose conditions on use and reuse of the research results they fund, at least with respect to literature and data.⁵¹⁷

For example, governments can dedicate government-generated work to the public domain, as occurs in the United States.⁵¹⁸ Funding agencies can mandate the deposit of publications in open access journals or, at least, in open access repositories,⁵¹⁹ as is happening ever more frequently in both the United States and the European Union. They can even impose analogs to fair use and to other codified limitations and exceptions by contract,⁵²⁰ which both publishers and individual scientists, as grantees, have to respect, especially if they wished to qualify for future grants. Funders can also support or reinforce self-archiving practices, and they increasingly provide for the costs of open access publishing in their grants.⁵²¹

Besides building open access provisions into their research grants, funding agencies can support the formation of digitally integrated research commons to serve the needs of diverse thematic communities.⁵²² Universities can lend their own weight to all these initiatives,⁵²³ and many have established open repositories for their employees’ scholarly works. Individual scientists can adopt existing Creative Commons and Science Commons licenses when publishing their works.⁵²⁴ Innovative proposals that go even farther, such as Victoria Stodden’s proposed Reproducible Research Standard, should also be tested and perfected.⁵²⁵

The common feature of these and other initiatives is that relevant information is made openly and freely available in digital format and online. Through many of these initiatives, material is made available either under suitably reduced proprietary terms and conditionsset out in permissive licenses⁵²⁶ (e.g., the GNU licenses for open source software,⁵²⁷ or Creative Commons licenses⁵²⁸ for open access journals or for some works in open repositories), or it will have entered the public domain.⁵²⁹ Under other mechanisms, such as the delayed open availability option, the works retain full copyright protection, but eventually become freely and openly accessible, at least on a read-only basis.⁵³⁰

Taken together, these activities are part of the emerging broader movement in support of both formal and informal peer production and dissemination of publicly funded scientific (and other) information in a globally distributed, voluntary, and open networked environment:⁵³¹

[They] are based on principles that reflect the cooperative ethos that traditionally has imbued much of [the] academic and government (civilian) research agencies; their norms and governance mechanisms may be characterized as those of “public scientific information commons,” rather than of a market system based upon proprietary data and information.⁵³²

How far these open access initiatives can be carried remains to be seen. The potential unwillingness of intermediaries or grantees to accept such contractual templates, in addition to intrinsic constraints on funders’ abilities to defray the costs of such institutional arrangements over time, effectively limit the regulatory powers of funders to achieve these objectives.⁵³³ With respect to grantees, a requirement to publish only in open access journals or only under Creative Commons or Science Commons licenses⁵³⁴ could hinder publication in some high prestige, peer-reviewed journals and breed resistance from leading members of the relevant scientific communities. Whether funding agencies, and the research community itself, can persuade these journals to become more open remains to be seen, but the evidence suggests that there is considerable momentum in that direction.⁵³⁵

C. Integrating the Intermediaries’ Functions into Transnational Digital Knowledge Environments

Aggressive resort to open access licensing conditions espoused by funders could, but not necessarily would, persuade some private publishers to abandon the field. This has not happened so far because funders are increasingly willing to enable grantees to purchase open access conditions from publishers at prices that appear to remain profitable for them. Pressure from funders can thus change the commercial publishers’ business model and persuade some to allow scientists to purchase open access rights and even make a profitable business out of selling such rights at about the same costs as publishing in an open access journal.⁵³⁶ Unfortunately, the percentage of grantees that actually opt to exercise this option, when not otherwise mandatory, still remains relatively small.⁵³⁷

Although reliance on intermediaries is deeply entrenched in the system, science policymakers might eventually want to reevaluate the costs and benefits of maintaining customary relationships with them and consider alternative strategies for disseminating research results. Such an exercise could, in particular, focus attention on the advantages of absorbing the publishing function, when feasible, into integrated, open-knowledge environments—as one study underway now advocates.⁵³⁸

Once anchored in appropriate institutions and freed from the legal and commercial fetters of both the professional societies and the commercial publishers, the very object of the publishing exercise could dramatically change. No longer would it be bound by obsolete concepts of the print model, which treat each monthly installment as a discrete legal and substantive unit. Rather, every new collection of research results made available to the relevant thematic community could enrich and expand an ever growing, digitally integrated database of aggregate scientific results.

Each of these thematically organized repositories, in turn, would remain fully open to data mining, manipulation, and other automated knowledge discovery tools, with full respect for reputational benefits but without palpable legal or economic constraints.⁵³⁹ Moreover, digital portals could link the formally published literature with so called grey literature, i.e., conference proceedings, and the like (which are not peer-reviewed). This aggregate resource can then be further linked with other data and relevant information bearing on all aspects of the science, including voluntarily contributed data pertaining to research of interest to a given thematic community.⁵⁴⁰

While this is not the place to fully elaborate on this concept, the astounding creative possibilities of unlimited, fully integrated knowledge hubs along these lines clearly dwarfs the gains that could be made from incremental or even structural reforms of the global intellectual property system. We believe that these or similar initiatives are essential for the progress of both science and culture, and would especially be needed to implement the sweeping new research vision that the National Research Council recently put forward for the life sciences.⁵⁴¹

Support for these and other initiatives could further encourage publishing intermediaries either to accommodate the open access movement or leave the scientific publishing business. By the same token, digitally integrated knowledge hubs could greatly magnify the creative and educational powers of universities and other analogous research institutions.⁵⁴²

For all these reasons, we question the customary practices of wholesale reliance on external-information brokers in a scientific world where it has become conceptually and technically feasible to link a given thematic community’s essential-knowledge resources into a seamless, digitally integrated network of inputs and outputs that remains open to all the contributors to any given research commons or semi-commons.⁵⁴³ The scientific community, now operating within a hostile intellectual property environment, faces the challenge of organizing and managing these knowledge assets with a view to establishing a broad upstream research space⁵⁴⁴ in which its own contractually imposed rules could apply without compromising the possibilities for commercial exploitation of downstream applications of the resulting research results.⁵⁴⁵

Nevertheless, the long-term drive to achieve these science policy goals should not obscure nor detract from the pressing short-term need to make the global intellectual property system more science-friendly than at present, along the lines we have explored. Legislatures concerned about the future of scientific research in the digital online environment⁵⁴⁶ should take steps now to reconfigure a legal domain that has become increasingly inimical to the needs of the scientific research community. Policymakers in OECD countries should join with key national institutions, such as the U.S. National Institutes of Health, in affirmatively promoting open access to scientific publications.

To this end, the relevant government agencies and private foundations should become funders of first resort for scientific publications and for the institutional repositories and e-commons in which those publications can be collected. Policymakers should likewise support the process of making government-funded research publications widely available through self-archiving and institutional archiving, with the fewest possible restrictions on use or reuse of published results.⁵⁴⁷

FINAL OBSERVATIONS

Scientific discoveries depend upon access to a robust public domain, in which preexisting discoveries become the building blocks of future investigations⁵⁴⁸ and existing information and data become inputs to future-knowledge assets that cannot be generated nearly as effectively without them.⁵⁴⁹ However, the recent tendency to elevate standards of intellectual property protection at both the national and international levels has been motivated largely by interests seeking to protect existing knowledge goods, destined mainly for end users, with insufficient regard for the social costs and burdens imposed on future creation and innovation, and with a corresponding bevy of new problems that hinder both objectives.⁵⁵⁰ This movement has generated thickets of intellectual property rights, high transaction and litigation costs, receding access to the public domain, growing anticommons effects, and the stifling of privileged uses by means of technological protection measures and digital rights management tools in the online environment.⁵⁵¹

In this Article, we have traced the contradictory measures in copyright and related laws that have increasingly impeded upstream scientific investigation and thereby complicated the exploitation of downstream applications of research results. By over-extending the protection of scientific information and data, these laws have made it harder for all investigators to build upon, rework, or further elaborate upon the contributions of others and to harness the astounding research potential of digital information technologies to their fullest extent.

From this perspective, the worldwide copyright system as it has lately evolved can hardly be said to benefit scientists qua authors. On the contrary, authors and compilers of scientific works and databases are still often obliged to surrender their outputs to publishers from whom they must buy back the very information and data they supplied (often at government expense). Rather than opening new vistas for producers of research data and information—as occurred after the printing press was invented and at regular intervals of technological change since then—copyright and database protection laws in the digitally networked environment seem bent on closing off new horizons in order to defend old business models for which publishers have sought few alternatives.⁵⁵²

A. Bridging the Disconnect Between Private Rights and Public Science

Given the opportunities that digital networks and automated knowledge discovery tools make possible, the logical goal for policymakers is to remove obstacles that the existing legal infrastructure poses for twenty-first century scientific endeavor. In this context, copyright law’s limitations and exceptions have an important role to play. They are not some nuisancelike sideshow of demands to be appeased as narrowly as possible. Rather, they should at least be viewed as a form of users’ rights,⁵⁵³ which help to supply inputs for scientific discoveries, innovation, and trade that are as indispensable to the dynamic production and dissemination of knowledge goods as suitably crafted incentives for authors and inventors.

A fundamental change of attitudes would be necessary. A top priority for policymakers should be to avoid generating legally established fiefdoms, in which a few private rights holders can combine the bulk of all scientific data and literature into monopolized repositories where access and use are restricted and controlled from the top down, and in which the commodified inputs of publicly funded science are distributed on a proprietary basis. Failure to achieve such a shift in priorities places digital and computational science in developed countries at risk of becoming progressively entangled in “copyright thickets”⁵⁵⁴ precisely at a time when these countries face stiff challenges from the growing scientific and technological capacities of the emerging economies.⁵⁵⁵

Despite the complexity of these issues, and the countervailing pressures of a powerful publishers’ lobby, policymakers need to resist the temptation to leave copyright and database protection laws where they stand or to strengthen them further in keeping with present trends. Few decisions could generate so many unintended harmful consequences. If these laws continue to impede e-science in the ways portrayed above, the much vaunted comparative advantages that industry and government spokespersons associate with maximalist levels of intellectual property protection could give way to private-sector strangleholds on the most promising avenues of public digital research, with the predictable result of killing the goose that lays the golden eggs.⁵⁵⁶

B. Reconciling the Goals of Innovation Policy with the Needs of Science Policy

In retrospect, it seems ironic that just as new technologies were producing significant breakthroughs in scientific research, and as digitally networked sites and other information technologies began empowering new models of collaborative investigation, innovation policies that should embrace these developments were instead using intellectual property rights to control or, in many cases, impede them. The successive use of public and private law to preclude access to basic knowledge resources, as well as knowledge-based goods, has increased the political and social burden of an intellectual property regime that, in theory, remains dedicated to the public interest of society at large.

Meanwhile, within the public-science community, efforts are underway to promote the formation of contractually constructed research commons (or semi-commons, as the case may be), that can flourish in an otherwise highly protectionist intellectual property environment. If successful, the resulting infrastructure could help to maintain a steady flow of downstream research products and socially beneficial commercial applications that do respond positively to the incentives of intellectual property rights.⁵⁵⁷ Given this transnational movement, what both the European Union and United States require is a long-term policy perspective that discriminates between the needs of the scientific community, operating within an emerging research commons that is increasingly capable of managing and integrating its own supplies of data and information, and the needs of the downstream technology sectors, which depend on the traditional incentives of intellectual property law to translate scientific discoveries into commercial applications.⁵⁵⁸

The object is to avoid pushing the exclusive rights that primarily govern those downstream incentives deep into the realm of basic science, where they will fracture and balkanize the research commons.⁵⁵⁹ Needed instead are measures that broaden the research commons and enable it to operate its computational tools in digitally integrated, field-specific communities that span the world, smoothly and without disruption from domestic toll collectors waiving IP stop signs.

These projects will require more than tinkering at the edges of copyright law. They will depend on some overall vision, a willingness to remove obstacles to modern research methods, and a determination to fund the necessary operations. Reforms on this scale will entail more than recognition of “users’ rights,”⁵⁶⁰ which denote important cultural interests and the public enrichment that ensues from access to literary and artistic works in general. Where science is concerned, information and data function as inputs to the process of discovery and thereby constitute an essential ingredient of future scientific progress.

Exclusive intellectual property rights do not provide the appropriate set of incentives in this upstream research space.⁵⁶¹ Policymakers should accordingly take pains to ensure that domestic and international intellectual property laws no longer undermine or impede the most promising opportunities that automated knowledge discover tools now make possible. These tools are critical for addressing the most pressing social and environmental challenges of our time.

Making the internet safe for publishers of print media should no longer justify hindering the aggregation of scientific information and data, or the uses of digitally integrated research methods capable of analyzing them on a global scale. Rather, the task is to reconcile the historical values of intellectual property law with the modalities of a digital age, in order to reinforce the needs of scientific investigators operating under twenty-first century conditions, and to stimulate maximum public welfare payoffs from their new technological tools.